diff --git a/.gitattributes b/.gitattributes index 5cc799d5b0780c0f83c1a1e5fe09b985964c3341..d7ab4e17227d7d85b0144def93777938316c3c40 100644 --- a/.gitattributes +++ b/.gitattributes @@ -5019,4 +5019,5 @@ lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbo lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-146.zst filter=lfs diff=lfs merge=lfs -text lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-183.zst filter=lfs diff=lfs merge=lfs -text lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-172.zst filter=lfs diff=lfs merge=lfs -text -lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst filter=lfs diff=lfs merge=lfs -text \ No newline at end of file +lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst filter=lfs diff=lfs merge=lfs -textlambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-188.zst filter=lfs diff=lfs merge=lfs -text +lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-20.zst filter=lfs diff=lfs merge=lfs -text diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5093.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5093.zst new file mode 100644 index 0000000000000000000000000000000000000000..4ee466bfe82bc7fc2d97c6cd1dd95b7372bda843 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5093.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8b39243af4e5e26011b189226290651aa7b9b05e015d1574b84530e08e435fa +size 1557064 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5118.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5118.zst new file mode 100644 index 0000000000000000000000000000000000000000..18e935bed96cc20e62fd513ceceebc0ac79d5dad --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5118.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5e2b5920864ad943365a95739592e89a4f82d39591a3286dedfbfda17f120c7 +size 1646379 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5144.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5144.zst new file mode 100644 index 0000000000000000000000000000000000000000..560fc8b5569e89b2aef9a22ecc869e6f083d9c02 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5144.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21b4ba2335432a5a79668e19dae1d8635d81dba00e674137de0eeb8521af3e54 +size 1507258 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5162.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5162.zst new file mode 100644 index 0000000000000000000000000000000000000000..1527015505adce71a1bf89da00fc2db51fae3871 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5162.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e0f9d95fa46eb4e4ea211a6ffc7cab8de42ea587688332eb1c8e1713de8aa7e +size 1679200 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5173.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5173.zst new file mode 100644 index 0000000000000000000000000000000000000000..c634ac9ed835d5a3a183721b7fef346274abec5c --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5173.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:297655759e27329d08f97958e6d8f1ba86e833ffc62d7bbd17cf36bf171cfa76 +size 1932362 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5186.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5186.zst new file mode 100644 index 0000000000000000000000000000000000000000..4513bb8750fc1901bf0fdc978b59e32dfc7be368 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5186.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89c72a40ed064cbb0e2c647d97d7a13738a396a91ded75c06612a4a293cd9032 +size 2112953 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5197.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5197.zst new file mode 100644 index 0000000000000000000000000000000000000000..9e8822496b1875c9423e1cd9bf88c76d0c489b04 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5197.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16692d0962b3a2194cb070ba33f5c9f375b29e7226fdfd4710a15a714b730534 +size 3672513 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5199.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5199.zst new file mode 100644 index 0000000000000000000000000000000000000000..99d04934173ecc8f0199a0dc4d1b5b745f03ef28 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5199.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89d1d73ffbb3f6518fa5ac377e60a3033e825180d7bfb8cac2c1c6ee38ff597c +size 3806032 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5204.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5204.zst new file mode 100644 index 0000000000000000000000000000000000000000..a5da6465aac5019cf69f76f82c7eebda2dde98f7 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5204.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14d034e53b3e5a2be9a6e263f962b289bd292f29378b0b6084677a4215b19961 +size 2669481 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5213.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5213.zst new file mode 100644 index 0000000000000000000000000000000000000000..4ff89a6ac4f10d9da7983d8a7641ba6e719b40f7 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5213.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:332f5eef1b442b7bf6c48b26c4f2f1db69f00c110adb1058adadb051f8aa6168 +size 2255574 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5216.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5216.zst new file mode 100644 index 0000000000000000000000000000000000000000..e4d56d434604a0038a253c761b099511613f3cf4 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5216.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f87aa3b1e28cded501b1eed8c36bb38318634850fe46303e536f6a80a2a18211 +size 2015657 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5218.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5218.zst new file mode 100644 index 0000000000000000000000000000000000000000..9cb8f6cf439294c03842c8811a532847e0520f12 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5218.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9378ee0d429c9ad2a6de54c70006d4a2a5d5e60ca255a605267828e2614225c +size 1991483 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5357.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5357.zst new file mode 100644 index 0000000000000000000000000000000000000000..70cb89728446cbc18a4c95f1d927e570288c91eb --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5357.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f193fe4cacee342d2d821a81ef70a8e6de2656429f5d90bb54b7d996acb688b3 +size 1228005 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5377.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5377.zst new file mode 100644 index 0000000000000000000000000000000000000000..98ba672b7590ab105bb89f2f9f131442a890d9ba --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5377.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e4a43d8797aa370995c0b57f0510619d72bfcbcc438d8be841ccb25152c539e +size 1455687 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5380.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5380.zst new file mode 100644 index 0000000000000000000000000000000000000000..549d1567bae39d6b5faa02750695f4b4f9ae9f55 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5380.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:62e1cbcb9eaf424cbbeef00d26dce3fc3edddd81718afd90de7d09fbd1a926f5 +size 1384808 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5397.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5397.zst new file mode 100644 index 0000000000000000000000000000000000000000..5af8c8986de7d52efa836e49b79f6aa2f67c4bef --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5397.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7424f170a2a535492ac29f9dc2183da97999df52888280b9fc2d46d1cc45522d +size 1534422 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5407.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5407.zst new file mode 100644 index 0000000000000000000000000000000000000000..9e40bb43f920a3f664c42b390039db82a8818995 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5407.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f21f9cd7d83317a76e372cc69aa81c91d03999e3a30dd57a5227cf056f9328b9 +size 1567037 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5423.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5423.zst new file mode 100644 index 0000000000000000000000000000000000000000..9ebf952c4f91b5ce4b311bae1241dde74195a5d8 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5423.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a742771b0988f50177d4ef00c7d97da74aac88036c90fe3002079a4e6e289c8 +size 1361339 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5430.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5430.zst new file mode 100644 index 0000000000000000000000000000000000000000..efc3c2b22e9bf1899bc8fda650c0aaca425f7ad7 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5430.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df6c55cb6a1c9308ea091ba63410df4db5f4260b1d266a3b9bc6ad21ec0c039b +size 1583662 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5437.zst b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5437.zst new file mode 100644 index 0000000000000000000000000000000000000000..ffe335ef84f40e7162718850b5e83a98a29847c5 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/librispeech-test-other/LibriSpeech-test_other-5437.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:76acb2e0a3654761485009fdb4557c4f99ab776b5f97178961dc02865ddbf320 +size 1418139 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-105.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-105.zst new file mode 100644 index 0000000000000000000000000000000000000000..49d720bbb39f75dbe976d51339fa5d40bae941a5 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-105.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03081a65438211a670d12cff45dd156f0f3cb1ce5b52ad7cbadaa3f427dafc2e +size 2371145 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-106.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-106.zst new file mode 100644 index 0000000000000000000000000000000000000000..11d672048d71c688fccd8353c34e0809ebdce564 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-106.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64e3484e27dbd6f9c4abee7a89da8077e9708c59c3752879def2b219d74d067e +size 2446456 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-108.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-108.zst new file mode 100644 index 0000000000000000000000000000000000000000..8dce5f2fa7f3773b4c464dbad8a2cdc90257ee14 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-108.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db3ef6243061432e791ac733938651cf7366956b0929aeacfe5f06eb1a24fe79 +size 2477795 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-110.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-110.zst new file mode 100644 index 0000000000000000000000000000000000000000..905960706fec3613b438121e0b20ab8651f8a1d0 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-110.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0bf3a1f697a5c09dd660125727106ec1d7b17849e062fd67eb90a3b0b4c3f3a +size 2345284 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-111.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-111.zst new file mode 100644 index 0000000000000000000000000000000000000000..0da417f75b41a28e89de9168fb7fc873ceb9b1cd --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-111.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cc7374aba4fd0eb1a3682e57f6877849f1f2cb742709c47de49f8a0321c729f +size 2385124 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-12.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-12.zst new file mode 100644 index 0000000000000000000000000000000000000000..72bf4842fbbf8188145fa2ceb015d15cc77c7ff2 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-12.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c83b8729a86284f5f4d42f5e1243b9fe5714b56eb8897dfa80460e34fd74749b +size 2447071 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-120.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-120.zst new file mode 100644 index 0000000000000000000000000000000000000000..d798e4d24f1ff9769b7844c103df8e06326981da --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-120.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71fc4fab30d4c946fa6525a62605958ca758d88f0b30ac4baae55a2c08b79dc6 +size 2468700 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-129.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-129.zst new file mode 100644 index 0000000000000000000000000000000000000000..0d10bb87ef0e6318f68f123543604557a6bca36f --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-129.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32b8222b548ce3d293eb88a72f0f70798213313d63c638196bb53dfe126cdd24 +size 2372923 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-13.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-13.zst new file mode 100644 index 0000000000000000000000000000000000000000..316b66291a60148d1ac7e7419d6f716186666b57 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-13.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bce47b61a31092b9356a6ac11ba5fc50d65083975b4bc8c31ed5e7a6f688f397 +size 2550216 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-133.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-133.zst new file mode 100644 index 0000000000000000000000000000000000000000..c519043ba0bcdcb25d75a3a0f52f11357158bf64 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-133.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65b91696e512c78b3967436651b8d072d08d2696e958316ca2d3fcd09783218a +size 2203700 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-135.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-135.zst new file mode 100644 index 0000000000000000000000000000000000000000..d1e04e63816c78d9ae546c3b9edf0da067641988 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-135.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b026bf4aef60b0f4d0cf05f0ca1c726d8dc3deab512e664d651ec49bebe68dcf +size 2479616 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-143.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-143.zst new file mode 100644 index 0000000000000000000000000000000000000000..5221e6ed64c37aa0395ce89c414242cb331cf00f --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-143.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4981fb917c651565d3c948439608393a93f98c01c6ad2e21b403d90ae74df701 +size 2528345 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-146.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-146.zst new file mode 100644 index 0000000000000000000000000000000000000000..03112ef3483e4d55fef037797eef2c6e3f8d173f --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-146.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58aeed0173a3b0bceb81da98d0b36aabc8456e48608896a97a9327fdb5396ad2 +size 2503178 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-148.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-148.zst new file mode 100644 index 0000000000000000000000000000000000000000..c66abb142e528f94ad3d9f7d512589662bc70a7d --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-148.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10b98a86ce765b42d0d89ad5d7a285734eae1735f5d3348454e5b7bdd663d004 +size 2288632 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-151.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-151.zst new file mode 100644 index 0000000000000000000000000000000000000000..8c9f452a0901dfcad299f8fa98367afd727c69b2 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-151.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96850ca69f2a10f950f221802eb9029c5e405b0adc03bc3f0266dcb99d7386cc +size 2355329 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-160.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-160.zst new file mode 100644 index 0000000000000000000000000000000000000000..c803b72b66fc77b5843bb5bbbf728169e0ec8687 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-160.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fff71141836c23b836f8ba5ca696d37d9df8d99a34691e24f8bc5277778c241c +size 2340790 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst new file mode 100644 index 0000000000000000000000000000000000000000..37a98d27b904f10ecbf276f00ed2495b9dd81d7a --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d059e8fc26dcc38fd08fe15802fd3c235ecbd9769ed30414ad90442deb5ba97 +size 2479578 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-164.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-164.zst new file mode 100644 index 0000000000000000000000000000000000000000..8a747808719fcc4033ee2e518fb4f9ed7422488c --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-164.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:922a0c76956167f5a53ffbd4deb4d369af716074f92afd565569cf68fc358df8 +size 2478701 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-172.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-172.zst new file mode 100644 index 0000000000000000000000000000000000000000..e30e056d052b613a5afc24f48fd34fbf3bcc7997 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-172.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd500f6b5f9b260658cf09825535cf08cdea96b59d99c2403d06f12a6a802a07 +size 2177564 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-174.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-174.zst new file mode 100644 index 0000000000000000000000000000000000000000..2c55d24cc889de6adec3a300feb881c94f1b8521 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-174.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7dee5a38511ef395df20ea1312d3eddbd1ac81f40cf74860f734a50321af207 +size 2181388 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-183.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-183.zst new file mode 100644 index 0000000000000000000000000000000000000000..2c8c01f014ade6ceba8785c8b6dc4d5b6a054281 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-183.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68f48ffe11532bd71acee4ad8867937449441037439be9fefba05cd3896cb569 +size 2336899 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-186.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-186.zst new file mode 100644 index 0000000000000000000000000000000000000000..e54849b5b03961c593ed8f9149693a33f5a78bc2 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-186.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0401b3e50141d601e472ee89db5fb679be4d8f812dee14ca0b9da7a7b3177be1 +size 2632449 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-188.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-188.zst new file mode 100644 index 0000000000000000000000000000000000000000..4cd2b09804e74eb5b387895d214b1c933ead6668 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-188.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:732fa1d6aeb9567e160108c2e874df3d9fff27b86963d4dc06a1b1ebff804a70 +size 2236860 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-194.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-194.zst new file mode 100644 index 0000000000000000000000000000000000000000..785c1941da835fd1c36576b7bfc2b303c9ebddba --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-194.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a8bcdfbebf05a59d04558d3bbf9f6ea4fd454c004752cf455e287dd3b8647b9 +size 2343786 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-20.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-20.zst new file mode 100644 index 0000000000000000000000000000000000000000..77dfe6d52bacd2ef4ab3a11cd8f7ff758aa5ed28 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-20.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e84d06d331c3d125ec10e10ff82f9d7dc2d0ff58a38437a7e79231ee2d22b84 +size 2269016 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-204.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-204.zst new file mode 100644 index 0000000000000000000000000000000000000000..91dbd3a3a7f3dc25bdcea71d27dbf20788227079 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-204.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20214e4f6d2a567710eef94febcf3b055e240d193118ef9e8586cdbb607f4af8 +size 2612792 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-207.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-207.zst new file mode 100644 index 0000000000000000000000000000000000000000..09e3178a038f78c1ae6613324e21683a886e1b23 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-207.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e8c6345f095e750cc67e3defd15683606fa71811812021c7de07c497197e1e0 +size 2217710 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-211.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-211.zst new file mode 100644 index 0000000000000000000000000000000000000000..f5c4f5c1346bb1b1b09b2f205676c1634c746d72 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-211.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a01026d89227bd45162924b85302ce3b87e924d911660dc123c062dfd8dcfef8 +size 2807635 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-215.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-215.zst new file mode 100644 index 0000000000000000000000000000000000000000..71c30d85f85d33f8624bd99253f209cdd2636aa9 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-215.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3062067146425b1c66b0b2574fb7b4a01e889052cc0cf4f42464e951a2a8be2e +size 2293812 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-216.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-216.zst new file mode 100644 index 0000000000000000000000000000000000000000..30bfb9327f3b2ef45ee66d49bab19cdab6641bd2 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-216.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5aca6710715f9cc565108f617e6ac49346528341082a3e39babd9fcd5b6570ab +size 2297410 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-22.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-22.zst new file mode 100644 index 0000000000000000000000000000000000000000..5bbf394cd776d5a287e250f416ebbb0b8a48e212 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-22.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1425b7807f4886d9cdf540df301838f59b43b0dc0e1115e8b4e1d0ba3d402824 +size 2266014 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-221.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-221.zst new file mode 100644 index 0000000000000000000000000000000000000000..42b2b972b8663a0bb9cda6ba3d5df6459214ae94 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-221.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fbc8a93821eb5f6b40a55fcb3a03342c9aec36a436c0b6ba09c2088b10912ba +size 2278602 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-222.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-222.zst new file mode 100644 index 0000000000000000000000000000000000000000..752e58eb11cdcef1d8433699c7ba6121f4ad0305 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-222.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2488e4a52dceca346da4755b29637cea7f55d07904fa6c7383b4eb15ee56b3e9 +size 2577455 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-223.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-223.zst new file mode 100644 index 0000000000000000000000000000000000000000..2e2a3e157fcb0a9126ebd5bd30cfa3dc6660cf65 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-223.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11ed5cfe6f4d71c44b75e6b3060a5c68545f9110e29a84971153adc6bdab4a68 +size 2332840 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-23.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-23.zst new file mode 100644 index 0000000000000000000000000000000000000000..3d6743a5b57672e25c5da40c3a794961d8975e82 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-23.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:828ed96274b55641bdfc07644c0303400475671f0fb7ee55bc9f0f36e3bb1063 +size 2324989 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-232.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-232.zst new file mode 100644 index 0000000000000000000000000000000000000000..9416d9c621e5ae564dfad214aa358d11d0ca2a22 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-232.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0226524a43655292e505bc92e59fe91e837ec8ed46bfe456216cc7d9fbd2536 +size 2367338 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-236.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-236.zst new file mode 100644 index 0000000000000000000000000000000000000000..f942db923af84153091c8a886c72df2c6cf568c2 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-236.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb761458a2fa6fd48ba68fff6353d8b93be6d90ca8dcc89ae2f3def08b0e7600 +size 2300569 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-24.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-24.zst new file mode 100644 index 0000000000000000000000000000000000000000..c40f284229c5b63b14902ce12f3ac6e5fc87e8f5 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-24.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc5d5e0539defcfefcdfe3c12176073293e0e04c95af96fe1cd183605405f75a +size 2343671 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-245.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-245.zst new file mode 100644 index 0000000000000000000000000000000000000000..bac0684efaa3bed295d2c4f52ba8f5791d75da0f --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-245.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aee57eb671ebc8ca8d24807986cfb555edf975355cc701e97766ebf25795de36 +size 2136998 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-246.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-246.zst new file mode 100644 index 0000000000000000000000000000000000000000..e74240c543b6f8c55a77d961c3010ce699bb71d7 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-246.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89e49ea5f05929eb1470bdbf68d9c72cdec135bf3445ab2ff2714890030e431b +size 2266340 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-249.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-249.zst new file mode 100644 index 0000000000000000000000000000000000000000..48d1a95e2d898b3f23e74e6c66f55128e180ac2b --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-249.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab3957ea9617fa90aaeea7e142a1e6de5bf1c1403e02c17e14681e29e065825f +size 2201427 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-250.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-250.zst new file mode 100644 index 0000000000000000000000000000000000000000..864152df9ac08ed1c5c6a0d110e1923853d21050 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-250.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8354739b76efa88eb1d0004b9a4c992d23b2e3390a1609e7ff389f272cf0186 +size 2514665 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-252.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-252.zst new file mode 100644 index 0000000000000000000000000000000000000000..5b36d8a8311a419871630e083fb4f0cc5a67c6af --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-252.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2beeff79925a6b2635f3ec838a9604d5be1b0551ea89a7918aecd6a468e7410b +size 2337933 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-253.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-253.zst new file mode 100644 index 0000000000000000000000000000000000000000..9218ffaa29c769fd41dfaf71e11fb67abdeb1768 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-253.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0982b47d48d97f9c515aa8ffaf1f62e3c81ae9240208a4667cbd34e9bb217c6 +size 2272333 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-254.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-254.zst new file mode 100644 index 0000000000000000000000000000000000000000..db32c2ea40d0a8ce11dd31528cef1a35ac98de51 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-254.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ba98a3246ba91eb8e67c08edc471c725d7c52f595facd1694beb3552da06bd6 +size 2355417 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-255.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-255.zst new file mode 100644 index 0000000000000000000000000000000000000000..0595277fdfd6d69ae830fd56140246c208695208 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-255.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f08ad11f53ea6eeb19fbacb250f1fc42216ccf45baf55a454acd21f2abf22b74 +size 2305484 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-258.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-258.zst new file mode 100644 index 0000000000000000000000000000000000000000..13362cce1b88c8ee917258c89b1245f33e0182ff --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-258.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:501932da0c116f3d391f0bef863a8cd33c3eedf5faf6349dbf615835de4e6ad5 +size 2384114 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-260.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-260.zst new file mode 100644 index 0000000000000000000000000000000000000000..6e4c73d4fe7133e31a25541c783fd23bdc3bdcda --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-260.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b135197aaca3828a0af2145d5f465378d6d4a0f982dabf193b268f29f26b1ef4 +size 2427311 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-264.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-264.zst new file mode 100644 index 0000000000000000000000000000000000000000..0cd8540d4217ce20220dcd6b3a21520d0421925c --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-264.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b64d8062acb727242334eba70e07ee6e4e87956b717b24b66b4bf3fef3dca01b +size 2447636 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-265.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-265.zst new file mode 100644 index 0000000000000000000000000000000000000000..677d84928d8ddf33dd2cea498f2cf87493bf8376 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-265.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92dc10d1b1a45a66599bf0b70a0d87fbd46cf5adbe5eaa04c9cb9295c2e8ffb0 +size 2489195 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-266.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-266.zst new file mode 100644 index 0000000000000000000000000000000000000000..5f968bea72625a7eaffd607e7fd7b188b3846389 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-266.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5b6992502db06187cc6fc45cc3bfc77481aa6ed05fd32a543184cd51dee1719 +size 2283060 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-267.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-267.zst new file mode 100644 index 0000000000000000000000000000000000000000..01c8f2266c0265d4060e71c0404aa50f9d2751a4 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-267.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4546d991b89efe6c91888099a726f300654cbe9343be6a0220adc4bb1b4e0db +size 2327109 diff --git a/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-269.zst b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-269.zst new file mode 100644 index 0000000000000000000000000000000000000000..440888856fab5a85aac25ab1880f99aeb5254418 --- /dev/null +++ b/lambda0.004/elic-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-269.zst @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:376415a4f840998e708be64007c4a9644a4864d381ebd945c0d31bfcbadedbab +size 2480751 diff --git a/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log b/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log new file mode 100644 index 0000000000000000000000000000000000000000..a8a9791073f62e825951d357f55e37f245bf3ed4 --- /dev/null +++ b/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log @@ -0,0 +1,14258 @@ +Experiment: dtufc_hyperprior-featurecoding_kimiaudio_individual +Log file: output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log +DTUFCCodecConfig: + arch: hyperprior-featurecoding + handler: kimiaudio + checkpoint: codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar + transform_type: kmeans + transform_mapping:featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json + bit_depth: 8 + device: cuda:0 +Loading checkpoint: codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar +Checkpoint epoch: 598 +Loaded hyperprior-featurecoding (1-channel) on cuda:0 +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_feature.json: torch.Size([256]) +Loaded per-key quantization points for key 'output' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_feature.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.0.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.0.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.1.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.1.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.2.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.2.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.3.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.3.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.4.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.4.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_v.json +Loaded per-key mappings: model=kimiaudio + Keys: ['output', 'layer.0.k_cache', 'layer.0.v_cache', 'layer.1.k_cache', 'layer.1.v_cache', 'layer.2.k_cache', 'layer.2.v_cache', 'layer.3.k_cache', 'layer.3.v_cache', 'layer.4.k_cache', 'layer.4.v_cache'] +---------------- ----------------------------------------------------------------------------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +Checkpoint codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar +Transform type kmeans +Transform config featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json +Input ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa +Output output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa +---------------- ----------------------------------------------------------------------------------------------------------------------------- +Files found: 100 +---------------------------------------------------------------------- + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-105.zst (1/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-105.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.012s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 288, 128) +Output shape: (1, 288, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.output: torch.Size([1, 288, 3584]) -> torch.Size([1, 1, 288, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,716B, BPFP=0.7984 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,252B, BPFP=2.6178 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,056B, BPFP=1.4136 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,960B, BPFP=2.5477 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,500B, BPFP=1.6547 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,320B, BPFP=2.5130 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,828B, BPFP=1.5640 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,980B, BPFP=2.5488 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,632B, BPFP=2.2044 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,452B, BPFP=2.4659 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,408B, BPFP=0.7627 +⌛️ [2/4] FRONTEND: Frontend time: 0.869s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.522s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14134069 22.52495660 + layer.0.v_cache 0.00001405 0.00501382 + layer.1.k_cache 0.62039534 3.26441426 + layer.1.v_cache 0.00000582 0.00208711 + layer.2.k_cache 0.00676800 0.42005171 + layer.2.v_cache 0.00001893 0.00574985 + layer.3.k_cache 0.03896382 2.07759751 + layer.3.v_cache 0.00001942 0.00654169 + layer.4.k_cache 0.00069280 0.11774867 + layer.4.v_cache 0.00005238 0.01257137 + layer.4.output 0.00804578 187.31990947 + ------------------------------------------------------------------------------------- + TOTAL 0.05085834 78.80447641 + (elements=2,506,752) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2506752 +Total Bytes 473104 +BPFP 1.5099 bits/point +EBPFP 3.0197 equivalent bits/point +MSE 78.804476 +---------------------- -------------------------------------------------------- +Time: 1.403s Load: 0.012s, Pack+Encode: 0.869s, Decode+Unpack: 0.522s +---------------------- -------------------------------------------------------- +💾 Converting with 78.8045 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-105.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-105.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-106.zst (2/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-106.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 293, 128) +Output shape: (1, 293, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.output: torch.Size([1, 293, 3584]) -> torch.Size([1, 1, 293, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,312B, BPFP=0.8166 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,632B, BPFP=2.6468 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,000B, BPFP=1.4398 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,460B, BPFP=2.5843 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,496B, BPFP=1.6796 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,752B, BPFP=2.5465 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,836B, BPFP=1.5911 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,524B, BPFP=2.5877 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,816B, BPFP=2.2299 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,912B, BPFP=2.5017 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,408B, BPFP=0.7878 +⌛️ [2/4] FRONTEND: Frontend time: 0.377s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.output: torch.Size([1, 293, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.532s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.output: torch.Size([1, 293, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15682776 22.59141025 + layer.0.v_cache 0.00001501 0.00512155 + layer.1.k_cache 0.54447479 3.04611300 + layer.1.v_cache 0.00000580 0.00224029 + layer.2.k_cache 0.00853899 0.44250671 + layer.2.v_cache 0.00001972 0.00581004 + layer.3.k_cache 0.02401569 1.91968190 + layer.3.v_cache 0.00001939 0.00635539 + layer.4.k_cache 0.00071059 0.11863283 + layer.4.v_cache 0.00005329 0.01268298 + layer.4.output 0.05121508 183.08614700 + ------------------------------------------------------------------------------------- + TOTAL 0.06430510 77.04432847 + (elements=2,550,272) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2550272 +Total Bytes 490148 +BPFP 1.5376 bits/point +EBPFP 3.0751 equivalent bits/point +MSE 77.044328 +---------------------- -------------------------------------------------------- +Time: 0.919s Load: 0.010s, Pack+Encode: 0.377s, Decode+Unpack: 0.532s +---------------------- -------------------------------------------------------- +💾 Converting with 77.0443 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-106.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-106.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-108.zst (3/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-108.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.012s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,408B, BPFP=0.8189 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,660B, BPFP=2.6392 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,992B, BPFP=1.4345 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,552B, BPFP=2.5804 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,348B, BPFP=1.6660 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,728B, BPFP=2.5366 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,600B, BPFP=1.5731 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,496B, BPFP=2.5774 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,492B, BPFP=2.2051 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,700B, BPFP=2.4819 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,052B, BPFP=0.7824 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.491s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15363285 22.60748831 + layer.0.v_cache 0.00001412 0.00489560 + layer.1.k_cache 0.60852461 3.07163348 + layer.1.v_cache 0.00000580 0.00204487 + layer.2.k_cache 0.01182756 0.44838149 + layer.2.v_cache 0.00001863 0.00549900 + layer.3.k_cache 0.03467803 1.94851560 + layer.3.v_cache 0.00001919 0.00646273 + layer.4.k_cache 0.00072319 0.11111328 + layer.4.v_cache 0.00005301 0.01194072 + layer.4.output 0.05035121 182.63509779 + ------------------------------------------------------------------------------------- + TOTAL 0.06835032 76.86256821 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 489028 +BPFP 1.5288 bits/point +EBPFP 3.0576 equivalent bits/point +MSE 76.862568 +---------------------- -------------------------------------------------------- +Time: 0.855s Load: 0.012s, Pack+Encode: 0.352s, Decode+Unpack: 0.491s +---------------------- -------------------------------------------------------- +💾 Converting with 76.8626 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-108.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-108.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-110.zst (4/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-110.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,972B, BPFP=0.8266 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,896B, BPFP=2.6444 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,160B, BPFP=1.4443 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,660B, BPFP=2.5762 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,472B, BPFP=1.6824 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,192B, BPFP=2.5504 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,700B, BPFP=1.5846 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,740B, BPFP=2.5806 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,564B, BPFP=2.2396 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,448B, BPFP=2.5093 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,964B, BPFP=0.8200 +⌛️ [2/4] FRONTEND: Frontend time: 0.351s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.494s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12197856 23.06143027 + layer.0.v_cache 0.00001365 0.00492054 + layer.1.k_cache 0.62524759 3.06110762 + layer.1.v_cache 0.00000551 0.00199571 + layer.2.k_cache 0.01486346 0.43725653 + layer.2.v_cache 0.00001981 0.00583962 + layer.3.k_cache 0.07387851 1.93902890 + layer.3.v_cache 0.00001943 0.00642928 + layer.4.k_cache 0.00069560 0.11447443 + layer.4.v_cache 0.00005405 0.01304864 + layer.4.output 0.01098318 187.95183935 + ------------------------------------------------------------------------------------- + TOTAL 0.05374461 79.07696512 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 477768 +BPFP 1.5517 bits/point +EBPFP 3.1034 equivalent bits/point +MSE 79.076965 +---------------------- -------------------------------------------------------- +Time: 0.854s Load: 0.008s, Pack+Encode: 0.351s, Decode+Unpack: 0.494s +---------------------- -------------------------------------------------------- +💾 Converting with 79.0770 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-110.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-110.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-111.zst (5/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-111.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 285, 128) +Output shape: (1, 285, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.output: torch.Size([1, 285, 3584]) -> torch.Size([1, 1, 285, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,252B, BPFP=0.8362 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,128B, BPFP=2.6386 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,348B, BPFP=1.4445 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,792B, BPFP=2.5654 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,476B, BPFP=1.6708 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,096B, BPFP=2.5272 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,872B, BPFP=1.5829 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,804B, BPFP=2.5660 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,408B, BPFP=2.2154 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,276B, BPFP=2.4822 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,128B, BPFP=0.8155 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.502s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15362121 22.48130482 + layer.0.v_cache 0.00001420 0.00501310 + layer.1.k_cache 0.52952420 3.01546567 + layer.1.v_cache 0.00000567 0.00206836 + layer.2.k_cache 0.00939641 0.43975260 + layer.2.v_cache 0.00001891 0.00559236 + layer.3.k_cache 0.04339437 1.95433735 + layer.3.v_cache 0.00002732 0.00643697 + layer.4.k_cache 0.00068326 0.11804922 + layer.4.v_cache 0.00005156 0.01250083 + layer.4.output 0.00776138 192.54244987 + ------------------------------------------------------------------------------------- + TOTAL 0.04653334 80.93162767 + (elements=2,480,640) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2480640 +Total Bytes 478580 +BPFP 1.5434 bits/point +EBPFP 3.0868 equivalent bits/point +MSE 80.931628 +---------------------- -------------------------------------------------------- +Time: 0.863s Load: 0.009s, Pack+Encode: 0.352s, Decode+Unpack: 0.502s +---------------------- -------------------------------------------------------- +💾 Converting with 80.9316 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-111.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-111.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-12.zst (6/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-12.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,048B, BPFP=0.7997 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,632B, BPFP=2.6378 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,968B, BPFP=1.4332 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,636B, BPFP=2.5848 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,440B, BPFP=1.6709 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,864B, BPFP=2.5438 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,356B, BPFP=1.5602 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,428B, BPFP=2.5738 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,852B, BPFP=2.2243 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,832B, BPFP=2.4889 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,180B, BPFP=0.7834 +⌛️ [2/4] FRONTEND: Frontend time: 0.347s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.496s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14921317 22.52594202 + layer.0.v_cache 0.00001395 0.00507422 + layer.1.k_cache 0.64083395 3.15245606 + layer.1.v_cache 0.00000574 0.00211927 + layer.2.k_cache 0.00578804 0.43062176 + layer.2.v_cache 0.00001912 0.00596886 + layer.3.k_cache 0.02848778 2.03736742 + layer.3.v_cache 0.00001956 0.00666667 + layer.4.k_cache 0.00070939 0.11629399 + layer.4.v_cache 0.00005206 0.01263694 + layer.4.output 0.04916091 182.55981232 + ------------------------------------------------------------------------------------- + TOTAL 0.06878054 76.83610785 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 489236 +BPFP 1.5295 bits/point +EBPFP 3.0589 equivalent bits/point +MSE 76.836108 +---------------------- -------------------------------------------------------- +Time: 0.853s Load: 0.010s, Pack+Encode: 0.347s, Decode+Unpack: 0.496s +---------------------- -------------------------------------------------------- +💾 Converting with 76.8361 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-12.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-12.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-120.zst (7/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-120.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 293, 128) +Output shape: (1, 293, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) -> torch.Size([1, 1, 293, 512]) + layer.4.output: torch.Size([1, 293, 3584]) -> torch.Size([1, 1, 293, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,044B, BPFP=0.8023 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,592B, BPFP=2.6446 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,704B, BPFP=1.4241 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,312B, BPFP=2.5764 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,456B, BPFP=1.6775 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,716B, BPFP=2.5446 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,552B, BPFP=1.5759 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,380B, BPFP=2.5800 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,684B, BPFP=2.2229 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,660B, BPFP=2.4883 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,060B, BPFP=0.7851 +⌛️ [2/4] FRONTEND: Frontend time: 0.372s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.output: torch.Size([1, 293, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.523s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 293, 128]) + layer.0.v_cache: torch.Size([1, 4, 293, 128]) + layer.1.k_cache: torch.Size([1, 4, 293, 128]) + layer.1.v_cache: torch.Size([1, 4, 293, 128]) + layer.2.k_cache: torch.Size([1, 4, 293, 128]) + layer.2.v_cache: torch.Size([1, 4, 293, 128]) + layer.3.k_cache: torch.Size([1, 4, 293, 128]) + layer.3.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.k_cache: torch.Size([1, 4, 293, 128]) + layer.4.v_cache: torch.Size([1, 4, 293, 128]) + layer.4.output: torch.Size([1, 293, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12806672 22.86349123 + layer.0.v_cache 0.00001388 0.00491411 + layer.1.k_cache 0.56701790 3.09738607 + layer.1.v_cache 0.00000557 0.00199287 + layer.2.k_cache 0.01020011 0.45057637 + layer.2.v_cache 0.00001881 0.00528002 + layer.3.k_cache 0.04301871 1.98996607 + layer.3.v_cache 0.00002051 0.00637755 + layer.4.k_cache 0.00071755 0.11478666 + layer.4.v_cache 0.00005284 0.01184329 + layer.4.output 0.04940383 183.21785105 + ------------------------------------------------------------------------------------- + TOTAL 0.06440938 77.12185715 + (elements=2,550,272) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2550272 +Total Bytes 488160 +BPFP 1.5313 bits/point +EBPFP 3.0626 equivalent bits/point +MSE 77.121857 +---------------------- -------------------------------------------------------- +Time: 0.905s Load: 0.010s, Pack+Encode: 0.372s, Decode+Unpack: 0.523s +---------------------- -------------------------------------------------------- +💾 Converting with 77.1219 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-120.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-120.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-129.zst (8/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-129.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 285, 128) +Output shape: (1, 285, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.output: torch.Size([1, 285, 3584]) -> torch.Size([1, 1, 285, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,296B, BPFP=0.8386 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,100B, BPFP=2.6371 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,512B, BPFP=1.4535 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,860B, BPFP=2.5691 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,668B, BPFP=1.6814 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,316B, BPFP=2.5393 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,108B, BPFP=1.5958 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,992B, BPFP=2.5763 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,696B, BPFP=2.2311 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,536B, BPFP=2.4965 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,740B, BPFP=0.8047 +⌛️ [2/4] FRONTEND: Frontend time: 0.373s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.531s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12903612 22.27651282 + layer.0.v_cache 0.00001458 0.00503556 + layer.1.k_cache 0.59860808 3.03270392 + layer.1.v_cache 0.00000581 0.00201650 + layer.2.k_cache 0.00940504 0.46504645 + layer.2.v_cache 0.00001954 0.00584077 + layer.3.k_cache 0.02644065 1.99183628 + layer.3.v_cache 0.00002029 0.00654916 + layer.4.k_cache 0.00068693 0.11716598 + layer.4.v_cache 0.00005376 0.01267479 + layer.4.output 0.00862506 192.80697055 + ------------------------------------------------------------------------------------- + TOTAL 0.04850978 81.03318683 + (elements=2,480,640) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2480640 +Total Bytes 478824 +BPFP 1.5442 bits/point +EBPFP 3.0884 equivalent bits/point +MSE 81.033187 +---------------------- -------------------------------------------------------- +Time: 0.914s Load: 0.011s, Pack+Encode: 0.373s, Decode+Unpack: 0.531s +---------------------- -------------------------------------------------------- +💾 Converting with 81.0332 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-129.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-129.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-13.zst (9/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-13.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 308, 128) +Output shape: (1, 308, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.output: torch.Size([1, 308, 3584]) -> torch.Size([1, 1, 308, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,552B, BPFP=0.7890 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,732B, BPFP=2.5737 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,804B, BPFP=1.4105 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,524B, BPFP=2.5124 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,168B, BPFP=1.6319 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,944B, BPFP=2.4830 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,404B, BPFP=1.5424 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,592B, BPFP=2.5158 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,564B, BPFP=2.1593 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,908B, BPFP=2.4304 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 109,164B, BPFP=0.7911 +⌛️ [2/4] FRONTEND: Frontend time: 0.349s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.output: torch.Size([1, 308, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.520s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.output: torch.Size([1, 308, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16469599 21.69024103 + layer.0.v_cache 0.00001357 0.00486538 + layer.1.k_cache 0.70445866 2.78250320 + layer.1.v_cache 0.00000558 0.00208792 + layer.2.k_cache 0.01447187 0.43048472 + layer.2.v_cache 0.00001924 0.00541964 + layer.3.k_cache 0.02617162 1.83519438 + layer.3.v_cache 0.00001974 0.00636313 + layer.4.k_cache 0.00070375 0.11284192 + layer.4.v_cache 0.00005076 0.01188309 + layer.4.output 0.04814868 167.62284033 + ------------------------------------------------------------------------------------- + TOTAL 0.07339127 70.60245687 + (elements=2,680,832) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2680832 +Total Bytes 504356 +BPFP 1.5051 bits/point +EBPFP 3.0101 equivalent bits/point +MSE 70.602457 +---------------------- -------------------------------------------------------- +Time: 0.879s Load: 0.010s, Pack+Encode: 0.349s, Decode+Unpack: 0.520s +---------------------- -------------------------------------------------------- +💾 Converting with 70.6025 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-13.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-13.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-133.zst (10/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-133.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 269, 128) +Output shape: (1, 269, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.output: torch.Size([1, 269, 3584]) -> torch.Size([1, 1, 269, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,256B, BPFP=0.8281 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,588B, BPFP=2.7061 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,972B, BPFP=1.4505 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,372B, BPFP=2.6355 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,036B, BPFP=1.6866 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,628B, BPFP=2.5922 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,116B, BPFP=1.5750 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,188B, BPFP=2.6248 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,164B, BPFP=2.2749 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,844B, BPFP=2.5467 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 93,580B, BPFP=0.7765 +⌛️ [2/4] FRONTEND: Frontend time: 0.334s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.489s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12132761 23.29365852 + layer.0.v_cache 0.00001366 0.00494077 + layer.1.k_cache 0.57241889 3.23203660 + layer.1.v_cache 0.00000556 0.00195969 + layer.2.k_cache 0.00926234 0.43919426 + layer.2.v_cache 0.00002145 0.00540244 + layer.3.k_cache 0.03109839 2.04026051 + layer.3.v_cache 0.00002257 0.00615901 + layer.4.k_cache 0.00070701 0.11208431 + layer.4.v_cache 0.00004920 0.01133713 + layer.4.output 0.01031505 203.92923526 + ------------------------------------------------------------------------------------- + TOTAL 0.04747836 85.68539295 + (elements=2,341,376) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2341376 +Total Bytes 453744 +BPFP 1.5503 bits/point +EBPFP 3.1007 equivalent bits/point +MSE 85.685393 +---------------------- -------------------------------------------------------- +Time: 0.831s Load: 0.008s, Pack+Encode: 0.334s, Decode+Unpack: 0.489s +---------------------- -------------------------------------------------------- +💾 Converting with 85.6854 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-133.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-133.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-135.zst (11/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-135.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 298, 128) +Output shape: (1, 298, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.output: torch.Size([1, 298, 3584]) -> torch.Size([1, 1, 298, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,500B, BPFP=0.8127 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,068B, BPFP=2.6252 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,452B, BPFP=1.4394 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,680B, BPFP=2.5524 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,604B, BPFP=1.6571 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,112B, BPFP=2.5227 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,748B, BPFP=1.5598 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,608B, BPFP=2.5487 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,928B, BPFP=2.1984 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,068B, BPFP=2.4679 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,536B, BPFP=0.7605 +⌛️ [2/4] FRONTEND: Frontend time: 0.361s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.499s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14357947 22.92819152 + layer.0.v_cache 0.00001420 0.00515975 + layer.1.k_cache 0.63723908 3.06552452 + layer.1.v_cache 0.00000573 0.00211662 + layer.2.k_cache 0.00765126 0.42232170 + layer.2.v_cache 0.00002073 0.00554024 + layer.3.k_cache 0.02843528 1.82772131 + layer.3.v_cache 0.00002112 0.00659330 + layer.4.k_cache 0.00069171 0.11368206 + layer.4.v_cache 0.00005234 0.01221434 + layer.4.output 0.04871205 180.07506891 + ------------------------------------------------------------------------------------- + TOTAL 0.06815855 75.81850281 + (elements=2,593,792) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2593792 +Total Bytes 490304 +BPFP 1.5122 bits/point +EBPFP 3.0245 equivalent bits/point +MSE 75.818503 +---------------------- -------------------------------------------------------- +Time: 0.871s Load: 0.011s, Pack+Encode: 0.361s, Decode+Unpack: 0.499s +---------------------- -------------------------------------------------------- +💾 Converting with 75.8185 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-135.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-135.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-143.zst (12/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-143.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 305, 128) +Output shape: (1, 305, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.0.v_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.1.k_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.1.v_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.2.k_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.2.v_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.3.k_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.3.v_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.4.k_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.4.v_cache: torch.Size([1, 4, 305, 128]) -> torch.Size([1, 1, 305, 512]) + layer.4.output: torch.Size([1, 305, 3584]) -> torch.Size([1, 1, 305, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 16,012B, BPFP=0.8203 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,588B, BPFP=2.5916 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,740B, BPFP=1.4211 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,328B, BPFP=2.5270 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,120B, BPFP=1.6455 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,828B, BPFP=2.5014 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,372B, BPFP=1.5559 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,356B, BPFP=2.5285 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,748B, BPFP=2.1900 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,840B, BPFP=2.4508 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,200B, BPFP=0.7406 +⌛️ [2/4] FRONTEND: Frontend time: 0.374s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 305, 128]) + layer.0.v_cache: torch.Size([1, 4, 305, 128]) + layer.1.k_cache: torch.Size([1, 4, 305, 128]) + layer.1.v_cache: torch.Size([1, 4, 305, 128]) + layer.2.k_cache: torch.Size([1, 4, 305, 128]) + layer.2.v_cache: torch.Size([1, 4, 305, 128]) + layer.3.k_cache: torch.Size([1, 4, 305, 128]) + layer.3.v_cache: torch.Size([1, 4, 305, 128]) + layer.4.k_cache: torch.Size([1, 4, 305, 128]) + layer.4.v_cache: torch.Size([1, 4, 305, 128]) + layer.4.output: torch.Size([1, 305, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.466s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 305, 128]) + layer.0.v_cache: torch.Size([1, 4, 305, 128]) + layer.1.k_cache: torch.Size([1, 4, 305, 128]) + layer.1.v_cache: torch.Size([1, 4, 305, 128]) + layer.2.k_cache: torch.Size([1, 4, 305, 128]) + layer.2.v_cache: torch.Size([1, 4, 305, 128]) + layer.3.k_cache: torch.Size([1, 4, 305, 128]) + layer.3.v_cache: torch.Size([1, 4, 305, 128]) + layer.4.k_cache: torch.Size([1, 4, 305, 128]) + layer.4.v_cache: torch.Size([1, 4, 305, 128]) + layer.4.output: torch.Size([1, 305, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16857396 22.99793481 + layer.0.v_cache 0.00001478 0.00520730 + layer.1.k_cache 0.58716421 3.26244137 + layer.1.v_cache 0.00000600 0.00222106 + layer.2.k_cache 0.00827454 0.43056130 + layer.2.v_cache 0.00002177 0.00582679 + layer.3.k_cache 0.02531170 1.97279193 + layer.3.v_cache 0.00001983 0.00666074 + layer.4.k_cache 0.00069463 0.11576531 + layer.4.v_cache 0.00005316 0.01301429 + layer.4.output 0.04836898 174.91189988 + ------------------------------------------------------------------------------------- + TOTAL 0.06639514 73.71739553 + (elements=2,654,720) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2654720 +Total Bytes 496132 +BPFP 1.4951 bits/point +EBPFP 2.9902 equivalent bits/point +MSE 73.717396 +---------------------- -------------------------------------------------------- +Time: 0.850s Load: 0.009s, Pack+Encode: 0.374s, Decode+Unpack: 0.466s +---------------------- -------------------------------------------------------- +💾 Converting with 73.7174 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-143.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-143.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-146.zst (13/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-146.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 298, 128) +Output shape: (1, 298, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.output: torch.Size([1, 298, 3584]) -> torch.Size([1, 1, 298, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,600B, BPFP=0.8180 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,048B, BPFP=2.6242 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,300B, BPFP=1.4314 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,868B, BPFP=2.5623 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,648B, BPFP=1.6594 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,132B, BPFP=2.5237 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,876B, BPFP=1.5665 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,780B, BPFP=2.5577 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,124B, BPFP=2.2087 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,220B, BPFP=2.4759 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,568B, BPFP=0.7833 +⌛️ [2/4] FRONTEND: Frontend time: 0.338s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.508s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14028978 22.25584299 + layer.0.v_cache 0.00001397 0.00509308 + layer.1.k_cache 0.59966084 2.94896150 + layer.1.v_cache 0.00000578 0.00216057 + layer.2.k_cache 0.01765286 0.46130443 + layer.2.v_cache 0.00001884 0.00568447 + layer.3.k_cache 0.02653811 1.84666842 + layer.3.v_cache 0.00001935 0.00654491 + layer.4.k_cache 0.00071527 0.11864241 + layer.4.v_cache 0.00005402 0.01251768 + layer.4.output 0.04907703 179.34897831 + ------------------------------------------------------------------------------------- + TOTAL 0.06638282 75.47683933 + (elements=2,593,792) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2593792 +Total Bytes 494164 +BPFP 1.5241 bits/point +EBPFP 3.0483 equivalent bits/point +MSE 75.476839 +---------------------- -------------------------------------------------------- +Time: 0.855s Load: 0.009s, Pack+Encode: 0.338s, Decode+Unpack: 0.508s +---------------------- -------------------------------------------------------- +💾 Converting with 75.4768 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-146.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-146.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-148.zst (14/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-148.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 272, 128) +Output shape: (1, 272, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.output: torch.Size([1, 272, 3584]) -> torch.Size([1, 1, 272, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,984B, BPFP=0.8033 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,792B, BPFP=2.6880 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,768B, BPFP=1.4228 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,616B, BPFP=2.6204 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,368B, BPFP=1.6870 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,148B, BPFP=2.5935 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,412B, BPFP=1.5747 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,772B, BPFP=2.6294 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,216B, BPFP=2.2528 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,212B, BPFP=2.5398 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,600B, BPFP=0.8092 +⌛️ [2/4] FRONTEND: Frontend time: 0.386s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.520s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13255837 22.58807014 + layer.0.v_cache 0.00001394 0.00486131 + layer.1.k_cache 0.54921044 3.04279821 + layer.1.v_cache 0.00000578 0.00198654 + layer.2.k_cache 0.01292971 0.42521095 + layer.2.v_cache 0.00001901 0.00555168 + layer.3.k_cache 0.07190662 1.93235958 + layer.3.v_cache 0.00001944 0.00618914 + layer.4.k_cache 0.00068159 0.10981449 + layer.4.v_cache 0.00005125 0.01189833 + layer.4.output 0.00790709 198.40311515 + ------------------------------------------------------------------------------------- + TOTAL 0.04839681 83.35003214 + (elements=2,367,488) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2367488 +Total Bytes 460888 +BPFP 1.5574 bits/point +EBPFP 3.1148 equivalent bits/point +MSE 83.350032 +---------------------- -------------------------------------------------------- +Time: 0.916s Load: 0.009s, Pack+Encode: 0.386s, Decode+Unpack: 0.520s +---------------------- -------------------------------------------------------- +💾 Converting with 83.3500 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-148.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-148.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-151.zst (15/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-151.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 282, 128) +Output shape: (1, 282, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.output: torch.Size([1, 282, 3584]) -> torch.Size([1, 1, 282, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,924B, BPFP=0.8269 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,780B, BPFP=2.6474 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,868B, BPFP=1.4333 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,548B, BPFP=2.5791 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,512B, BPFP=1.6906 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,920B, BPFP=2.5443 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,680B, BPFP=1.5891 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,708B, BPFP=2.5880 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,508B, BPFP=2.2445 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,332B, BPFP=2.5117 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,392B, BPFP=0.8263 +⌛️ [2/4] FRONTEND: Frontend time: 0.499s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.output: torch.Size([1, 282, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.570s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.output: torch.Size([1, 282, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13695534 22.62731501 + layer.0.v_cache 0.00001452 0.00516909 + layer.1.k_cache 0.52272531 3.02066538 + layer.1.v_cache 0.00000574 0.00212650 + layer.2.k_cache 0.00793556 0.42606576 + layer.2.v_cache 0.00001907 0.00590271 + layer.3.k_cache 0.04682703 1.97210066 + layer.3.v_cache 0.00002008 0.00663889 + layer.4.k_cache 0.00068945 0.11895695 + layer.4.v_cache 0.00005221 0.01287278 + layer.4.output 0.00941004 194.25554078 + ------------------------------------------------------------------------------------- + TOTAL 0.04594792 81.64627054 + (elements=2,454,528) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2454528 +Total Bytes 477172 +BPFP 1.5552 bits/point +EBPFP 3.1105 equivalent bits/point +MSE 81.646271 +---------------------- -------------------------------------------------------- +Time: 1.080s Load: 0.011s, Pack+Encode: 0.499s, Decode+Unpack: 0.570s +---------------------- -------------------------------------------------------- +💾 Converting with 81.6463 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-151.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-151.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-160.zst (16/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-160.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 280, 128) +Output shape: (1, 280, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.0.v_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.1.k_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.1.v_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.2.k_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.2.v_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.3.k_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.3.v_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.4.k_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.4.v_cache: torch.Size([1, 4, 280, 128]) -> torch.Size([1, 1, 280, 512]) + layer.4.output: torch.Size([1, 280, 3584]) -> torch.Size([1, 1, 280, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,856B, BPFP=0.8290 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,448B, BPFP=2.6478 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,636B, BPFP=1.4306 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,264B, BPFP=2.5817 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,824B, BPFP=1.6643 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,596B, BPFP=2.5444 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,148B, BPFP=1.5708 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,316B, BPFP=2.5846 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,832B, BPFP=2.2228 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,868B, BPFP=2.5038 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,036B, BPFP=0.7895 +⌛️ [2/4] FRONTEND: Frontend time: 0.364s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 280, 128]) + layer.0.v_cache: torch.Size([1, 4, 280, 128]) + layer.1.k_cache: torch.Size([1, 4, 280, 128]) + layer.1.v_cache: torch.Size([1, 4, 280, 128]) + layer.2.k_cache: torch.Size([1, 4, 280, 128]) + layer.2.v_cache: torch.Size([1, 4, 280, 128]) + layer.3.k_cache: torch.Size([1, 4, 280, 128]) + layer.3.v_cache: torch.Size([1, 4, 280, 128]) + layer.4.k_cache: torch.Size([1, 4, 280, 128]) + layer.4.v_cache: torch.Size([1, 4, 280, 128]) + layer.4.output: torch.Size([1, 280, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.501s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 280, 128]) + layer.0.v_cache: torch.Size([1, 4, 280, 128]) + layer.1.k_cache: torch.Size([1, 4, 280, 128]) + layer.1.v_cache: torch.Size([1, 4, 280, 128]) + layer.2.k_cache: torch.Size([1, 4, 280, 128]) + layer.2.v_cache: torch.Size([1, 4, 280, 128]) + layer.3.k_cache: torch.Size([1, 4, 280, 128]) + layer.3.v_cache: torch.Size([1, 4, 280, 128]) + layer.4.k_cache: torch.Size([1, 4, 280, 128]) + layer.4.v_cache: torch.Size([1, 4, 280, 128]) + layer.4.output: torch.Size([1, 280, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14340984 22.63476562 + layer.0.v_cache 0.00001397 0.00488609 + layer.1.k_cache 0.51622396 3.17956979 + layer.1.v_cache 0.00000557 0.00201250 + layer.2.k_cache 0.00797994 0.40724133 + layer.2.v_cache 0.00001813 0.00520672 + layer.3.k_cache 0.02756977 1.86090829 + layer.3.v_cache 0.00001890 0.00610248 + layer.4.k_cache 0.00068956 0.11408872 + layer.4.v_cache 0.00005177 0.01195288 + layer.4.output 0.00947078 193.92857143 + ------------------------------------------------------------------------------------- + TOTAL 0.04483982 81.51333732 + (elements=2,437,120) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2437120 +Total Bytes 467824 +BPFP 1.5357 bits/point +EBPFP 3.0713 equivalent bits/point +MSE 81.513337 +---------------------- -------------------------------------------------------- +Time: 0.875s Load: 0.010s, Pack+Encode: 0.364s, Decode+Unpack: 0.501s +---------------------- -------------------------------------------------------- +💾 Converting with 81.5133 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-160.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-160.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-163.zst (17/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-163.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,004B, BPFP=0.7974 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,588B, BPFP=2.6354 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,088B, BPFP=1.4396 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,440B, BPFP=2.5744 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,364B, BPFP=1.6669 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,700B, BPFP=2.5351 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,532B, BPFP=1.5695 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,272B, BPFP=2.5655 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,692B, BPFP=2.2158 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,808B, BPFP=2.4877 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,524B, BPFP=0.7632 +⌛️ [2/4] FRONTEND: Frontend time: 0.371s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.503s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13622742 23.06295839 + layer.0.v_cache 0.00001380 0.00490014 + layer.1.k_cache 0.61520739 3.14348618 + layer.1.v_cache 0.00000584 0.00217099 + layer.2.k_cache 0.00900942 0.42574518 + layer.2.v_cache 0.00001916 0.00565804 + layer.3.k_cache 0.02667915 1.97283645 + layer.3.v_cache 0.00001881 0.00629157 + layer.4.k_cache 0.00071466 0.11594626 + layer.4.v_cache 0.00005760 0.01214283 + layer.4.output 0.04972813 182.67723518 + ------------------------------------------------------------------------------------- + TOTAL 0.06682648 76.91134013 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 486012 +BPFP 1.5194 bits/point +EBPFP 3.0388 equivalent bits/point +MSE 76.911340 +---------------------- -------------------------------------------------------- +Time: 0.885s Load: 0.010s, Pack+Encode: 0.371s, Decode+Unpack: 0.503s +---------------------- -------------------------------------------------------- +💾 Converting with 76.9113 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-163.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-163.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-164.zst (18/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-164.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 299, 128) +Output shape: (1, 299, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.output: torch.Size([1, 299, 3584]) -> torch.Size([1, 1, 299, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,532B, BPFP=0.8117 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,180B, BPFP=2.6223 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,500B, BPFP=1.4371 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,924B, BPFP=2.5566 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,728B, BPFP=1.6580 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,328B, BPFP=2.5255 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,036B, BPFP=1.5696 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,064B, BPFP=2.5640 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,112B, BPFP=2.2007 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,344B, BPFP=2.4741 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,940B, BPFP=0.7610 +⌛️ [2/4] FRONTEND: Frontend time: 0.351s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.output: torch.Size([1, 299, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.524s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.output: torch.Size([1, 299, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11478914 22.20222029 + layer.0.v_cache 0.00001377 0.00485786 + layer.1.k_cache 0.64218915 3.03576599 + layer.1.v_cache 0.00000556 0.00188453 + layer.2.k_cache 0.00898698 0.42701548 + layer.2.v_cache 0.00001980 0.00548203 + layer.3.k_cache 0.05442990 1.84405068 + layer.3.v_cache 0.00001964 0.00638058 + layer.4.k_cache 0.00070461 0.11118097 + layer.4.v_cache 0.00005015 0.01195293 + layer.4.output 0.05017195 174.99157907 + ------------------------------------------------------------------------------------- + TOTAL 0.06896543 73.68187323 + (elements=2,602,496) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2602496 +Total Bytes 492688 +BPFP 1.5145 bits/point +EBPFP 3.0290 equivalent bits/point +MSE 73.681873 +---------------------- -------------------------------------------------------- +Time: 0.884s Load: 0.009s, Pack+Encode: 0.351s, Decode+Unpack: 0.524s +---------------------- -------------------------------------------------------- +💾 Converting with 73.6819 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-164.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-164.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-172.zst (19/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-172.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 265, 128) +Output shape: (1, 265, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.0.v_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.1.k_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.1.v_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.2.k_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.2.v_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.3.k_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.3.v_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.4.k_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.4.v_cache: torch.Size([1, 4, 265, 128]) -> torch.Size([1, 1, 265, 512]) + layer.4.output: torch.Size([1, 265, 3584]) -> torch.Size([1, 1, 265, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,000B, BPFP=0.8255 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,148B, BPFP=2.7210 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,584B, BPFP=1.4495 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,120B, BPFP=2.6604 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,680B, BPFP=1.6910 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,412B, BPFP=2.6186 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,824B, BPFP=1.5816 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,048B, BPFP=2.6561 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,828B, BPFP=2.2894 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,568B, BPFP=2.5689 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 90,452B, BPFP=0.7619 +⌛️ [2/4] FRONTEND: Frontend time: 0.398s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 265, 128]) + layer.0.v_cache: torch.Size([1, 4, 265, 128]) + layer.1.k_cache: torch.Size([1, 4, 265, 128]) + layer.1.v_cache: torch.Size([1, 4, 265, 128]) + layer.2.k_cache: torch.Size([1, 4, 265, 128]) + layer.2.v_cache: torch.Size([1, 4, 265, 128]) + layer.3.k_cache: torch.Size([1, 4, 265, 128]) + layer.3.v_cache: torch.Size([1, 4, 265, 128]) + layer.4.k_cache: torch.Size([1, 4, 265, 128]) + layer.4.v_cache: torch.Size([1, 4, 265, 128]) + layer.4.output: torch.Size([1, 265, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.559s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 265, 128]) + layer.0.v_cache: torch.Size([1, 4, 265, 128]) + layer.1.k_cache: torch.Size([1, 4, 265, 128]) + layer.1.v_cache: torch.Size([1, 4, 265, 128]) + layer.2.k_cache: torch.Size([1, 4, 265, 128]) + layer.2.v_cache: torch.Size([1, 4, 265, 128]) + layer.3.k_cache: torch.Size([1, 4, 265, 128]) + layer.3.v_cache: torch.Size([1, 4, 265, 128]) + layer.4.k_cache: torch.Size([1, 4, 265, 128]) + layer.4.v_cache: torch.Size([1, 4, 265, 128]) + layer.4.output: torch.Size([1, 265, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14755792 22.96625516 + layer.0.v_cache 0.00001375 0.00513514 + layer.1.k_cache 0.51940866 3.03250111 + layer.1.v_cache 0.00000586 0.00219255 + layer.2.k_cache 0.00816260 0.43729795 + layer.2.v_cache 0.00001900 0.00583453 + layer.3.k_cache 0.02921419 1.95960210 + layer.3.v_cache 0.00001979 0.00676700 + layer.4.k_cache 0.00071017 0.11953705 + layer.4.v_cache 0.00005507 0.01255064 + layer.4.output 0.00937099 207.03357480 + ------------------------------------------------------------------------------------- + TOTAL 0.04533906 86.92839393 + (elements=2,306,560) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2306560 +Total Bytes 447664 +BPFP 1.5527 bits/point +EBPFP 3.1053 equivalent bits/point +MSE 86.928394 +---------------------- -------------------------------------------------------- +Time: 0.968s Load: 0.011s, Pack+Encode: 0.398s, Decode+Unpack: 0.559s +---------------------- -------------------------------------------------------- +💾 Converting with 86.9284 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-172.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-172.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-174.zst (20/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-174.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 267, 128) +Output shape: (1, 267, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.0.v_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.1.k_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.1.v_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.2.k_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.2.v_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.3.k_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.3.v_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.4.k_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.4.v_cache: torch.Size([1, 4, 267, 128]) -> torch.Size([1, 1, 267, 512]) + layer.4.output: torch.Size([1, 267, 3584]) -> torch.Size([1, 1, 267, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,120B, BPFP=0.8263 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,424B, BPFP=2.7168 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,752B, BPFP=1.4485 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,220B, BPFP=2.6463 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,968B, BPFP=1.6952 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,596B, BPFP=2.6098 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,188B, BPFP=1.5911 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,184B, BPFP=2.6442 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,188B, BPFP=2.2933 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,908B, BPFP=2.5695 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 94,336B, BPFP=0.7887 +⌛️ [2/4] FRONTEND: Frontend time: 0.387s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 267, 128]) + layer.0.v_cache: torch.Size([1, 4, 267, 128]) + layer.1.k_cache: torch.Size([1, 4, 267, 128]) + layer.1.v_cache: torch.Size([1, 4, 267, 128]) + layer.2.k_cache: torch.Size([1, 4, 267, 128]) + layer.2.v_cache: torch.Size([1, 4, 267, 128]) + layer.3.k_cache: torch.Size([1, 4, 267, 128]) + layer.3.v_cache: torch.Size([1, 4, 267, 128]) + layer.4.k_cache: torch.Size([1, 4, 267, 128]) + layer.4.v_cache: torch.Size([1, 4, 267, 128]) + layer.4.output: torch.Size([1, 267, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.538s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 267, 128]) + layer.0.v_cache: torch.Size([1, 4, 267, 128]) + layer.1.k_cache: torch.Size([1, 4, 267, 128]) + layer.1.v_cache: torch.Size([1, 4, 267, 128]) + layer.2.k_cache: torch.Size([1, 4, 267, 128]) + layer.2.v_cache: torch.Size([1, 4, 267, 128]) + layer.3.k_cache: torch.Size([1, 4, 267, 128]) + layer.3.v_cache: torch.Size([1, 4, 267, 128]) + layer.4.k_cache: torch.Size([1, 4, 267, 128]) + layer.4.v_cache: torch.Size([1, 4, 267, 128]) + layer.4.output: torch.Size([1, 267, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12840328 23.02825448 + layer.0.v_cache 0.00001372 0.00484281 + layer.1.k_cache 0.60081110 2.99549734 + layer.1.v_cache 0.00000562 0.00203940 + layer.2.k_cache 0.00586274 0.43512986 + layer.2.v_cache 0.00001839 0.00554504 + layer.3.k_cache 0.02491910 2.00604454 + layer.3.v_cache 0.00001957 0.00633712 + layer.4.k_cache 0.00072047 0.11907622 + layer.4.v_cache 0.00005229 0.01247861 + layer.4.output 0.01050496 200.78833935 + ------------------------------------------------------------------------------------- + TOTAL 0.04908006 84.36080123 + (elements=2,323,968) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2323968 +Total Bytes 453884 +BPFP 1.5624 bits/point +EBPFP 3.1249 equivalent bits/point +MSE 84.360801 +---------------------- -------------------------------------------------------- +Time: 0.934s Load: 0.009s, Pack+Encode: 0.387s, Decode+Unpack: 0.538s +---------------------- -------------------------------------------------------- +💾 Converting with 84.3608 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-174.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-174.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-183.zst (21/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-183.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,020B, BPFP=0.8293 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,796B, BPFP=2.6389 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,100B, BPFP=1.4410 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,328B, BPFP=2.5579 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,316B, BPFP=1.6738 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,912B, BPFP=2.5349 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,460B, BPFP=1.5713 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,528B, BPFP=2.5689 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,264B, BPFP=2.2231 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,108B, BPFP=2.4905 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,084B, BPFP=0.7973 +⌛️ [2/4] FRONTEND: Frontend time: 0.355s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.482s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14595435 22.86101066 + layer.0.v_cache 0.00001364 0.00468289 + layer.1.k_cache 0.59759904 2.88814240 + layer.1.v_cache 0.00000566 0.00195455 + layer.2.k_cache 0.01409306 0.42803788 + layer.2.v_cache 0.00002047 0.00544688 + layer.3.k_cache 0.06079736 2.06546420 + layer.3.v_cache 0.00002215 0.00619945 + layer.4.k_cache 0.00070776 0.11201331 + layer.4.v_cache 0.00005221 0.01220971 + layer.4.output 0.00769303 189.57993122 + ------------------------------------------------------------------------------------- + TOTAL 0.05135982 79.73204003 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 472916 +BPFP 1.5359 bits/point +EBPFP 3.0718 equivalent bits/point +MSE 79.732040 +---------------------- -------------------------------------------------------- +Time: 0.846s Load: 0.009s, Pack+Encode: 0.355s, Decode+Unpack: 0.482s +---------------------- -------------------------------------------------------- +💾 Converting with 79.7320 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-183.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-183.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-186.zst (22/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-186.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 315, 128) +Output shape: (1, 315, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.0.v_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.1.k_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.1.v_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.2.k_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.2.v_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.3.k_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.3.v_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.4.k_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.4.v_cache: torch.Size([1, 4, 315, 128]) -> torch.Size([1, 1, 315, 512]) + layer.4.output: torch.Size([1, 315, 3584]) -> torch.Size([1, 1, 315, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,728B, BPFP=0.7802 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 51,124B, BPFP=2.5359 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,780B, BPFP=1.3780 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,872B, BPFP=2.4738 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,440B, BPFP=1.6091 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 49,184B, BPFP=2.4397 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,660B, BPFP=1.5208 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,760B, BPFP=2.4683 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 43,020B, BPFP=2.1339 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 48,204B, BPFP=2.3911 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 110,812B, BPFP=0.7852 +⌛️ [2/4] FRONTEND: Frontend time: 0.365s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 315, 128]) + layer.0.v_cache: torch.Size([1, 4, 315, 128]) + layer.1.k_cache: torch.Size([1, 4, 315, 128]) + layer.1.v_cache: torch.Size([1, 4, 315, 128]) + layer.2.k_cache: torch.Size([1, 4, 315, 128]) + layer.2.v_cache: torch.Size([1, 4, 315, 128]) + layer.3.k_cache: torch.Size([1, 4, 315, 128]) + layer.3.v_cache: torch.Size([1, 4, 315, 128]) + layer.4.k_cache: torch.Size([1, 4, 315, 128]) + layer.4.v_cache: torch.Size([1, 4, 315, 128]) + layer.4.output: torch.Size([1, 315, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.505s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 315, 128]) + layer.0.v_cache: torch.Size([1, 4, 315, 128]) + layer.1.k_cache: torch.Size([1, 4, 315, 128]) + layer.1.v_cache: torch.Size([1, 4, 315, 128]) + layer.2.k_cache: torch.Size([1, 4, 315, 128]) + layer.2.v_cache: torch.Size([1, 4, 315, 128]) + layer.3.k_cache: torch.Size([1, 4, 315, 128]) + layer.3.v_cache: torch.Size([1, 4, 315, 128]) + layer.4.k_cache: torch.Size([1, 4, 315, 128]) + layer.4.v_cache: torch.Size([1, 4, 315, 128]) + layer.4.output: torch.Size([1, 315, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15500355 22.35840464 + layer.0.v_cache 0.00001366 0.00496467 + layer.1.k_cache 0.68167124 3.05402115 + layer.1.v_cache 0.00000571 0.00218428 + layer.2.k_cache 0.01007247 0.42830888 + layer.2.v_cache 0.00001936 0.00581879 + layer.3.k_cache 0.03272899 1.89911702 + layer.3.v_cache 0.00002006 0.00642824 + layer.4.k_cache 0.00071175 0.11792581 + layer.4.v_cache 0.00005427 0.01241210 + layer.4.output 0.04584261 166.87162698 + ------------------------------------------------------------------------------------- + TOTAL 0.07065879 70.35241026 + (elements=2,741,760) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2741760 +Total Bytes 508584 +BPFP 1.4840 bits/point +EBPFP 2.9679 equivalent bits/point +MSE 70.352410 +---------------------- -------------------------------------------------------- +Time: 0.880s Load: 0.011s, Pack+Encode: 0.365s, Decode+Unpack: 0.505s +---------------------- -------------------------------------------------------- +💾 Converting with 70.3524 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-186.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-186.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-188.zst (23/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-188.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 271, 128) +Output shape: (1, 271, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.output: torch.Size([1, 271, 3584]) -> torch.Size([1, 1, 271, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,220B, BPFP=0.8199 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,688B, BPFP=2.6919 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,828B, BPFP=1.4315 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,372B, BPFP=2.6160 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,124B, BPFP=1.6792 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,812B, BPFP=2.5837 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,156B, BPFP=1.5657 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,356B, BPFP=2.6151 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,120B, BPFP=2.2555 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,920B, BPFP=2.5323 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 91,952B, BPFP=0.7574 +⌛️ [2/4] FRONTEND: Frontend time: 0.344s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.output: torch.Size([1, 271, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.526s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.output: torch.Size([1, 271, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12889319 22.79644978 + layer.0.v_cache 0.00001350 0.00495195 + layer.1.k_cache 0.58775566 3.05058756 + layer.1.v_cache 0.00000553 0.00204745 + layer.2.k_cache 0.00723753 0.41646007 + layer.2.v_cache 0.00001924 0.00567915 + layer.3.k_cache 0.01761036 1.97102429 + layer.3.v_cache 0.00001847 0.00645171 + layer.4.k_cache 0.00071814 0.11269180 + layer.4.v_cache 0.00005270 0.01238831 + layer.4.output 0.01035140 198.53021218 + ------------------------------------------------------------------------------------- + TOTAL 0.04792848 83.41707161 + (elements=2,358,784) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2358784 +Total Bytes 452548 +BPFP 1.5349 bits/point +EBPFP 3.0697 equivalent bits/point +MSE 83.417072 +---------------------- -------------------------------------------------------- +Time: 0.879s Load: 0.009s, Pack+Encode: 0.344s, Decode+Unpack: 0.526s +---------------------- -------------------------------------------------------- +💾 Converting with 83.4171 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-188.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-188.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-194.zst (24/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-194.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 284, 128) +Output shape: (1, 284, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.output: torch.Size([1, 284, 3584]) -> torch.Size([1, 1, 284, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,964B, BPFP=0.8233 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,948B, BPFP=2.6380 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,968B, BPFP=1.4287 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,632B, BPFP=2.5656 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,240B, BPFP=1.6637 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,904B, BPFP=2.5255 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,392B, BPFP=1.5621 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,492B, BPFP=2.5579 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,472B, BPFP=2.2267 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,132B, BPFP=2.4831 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,612B, BPFP=0.7672 +⌛️ [2/4] FRONTEND: Frontend time: 0.334s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.output: torch.Size([1, 284, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.484s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.output: torch.Size([1, 284, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13033964 22.58003342 + layer.0.v_cache 0.00001368 0.00490663 + layer.1.k_cache 0.63029007 3.13616943 + layer.1.v_cache 0.00000551 0.00197057 + layer.2.k_cache 0.00637632 0.44860891 + layer.2.v_cache 0.00001960 0.00543247 + layer.3.k_cache 0.02368569 1.88812815 + layer.3.v_cache 0.00001974 0.00593284 + layer.4.k_cache 0.00068050 0.11284337 + layer.4.v_cache 0.00005209 0.01168844 + layer.4.output 0.00765416 190.84264965 + ------------------------------------------------------------------------------------- + TOTAL 0.04970952 80.24083893 + (elements=2,471,936) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2471936 +Total Bytes 469756 +BPFP 1.5203 bits/point +EBPFP 3.0406 equivalent bits/point +MSE 80.240839 +---------------------- -------------------------------------------------------- +Time: 0.827s Load: 0.010s, Pack+Encode: 0.334s, Decode+Unpack: 0.484s +---------------------- -------------------------------------------------------- +💾 Converting with 80.2408 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-194.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-194.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-20.zst (25/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-20.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 272, 128) +Output shape: (1, 272, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.output: torch.Size([1, 272, 3584]) -> torch.Size([1, 1, 272, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,188B, BPFP=0.8150 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,712B, BPFP=2.6834 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,884B, BPFP=1.4295 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,560B, BPFP=2.6172 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,260B, BPFP=1.6808 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,044B, BPFP=2.5875 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,456B, BPFP=1.5772 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,852B, BPFP=2.6340 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,296B, BPFP=2.2574 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,268B, BPFP=2.5430 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 95,908B, BPFP=0.7871 +⌛️ [2/4] FRONTEND: Frontend time: 0.342s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.502s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15644295 22.46882181 + layer.0.v_cache 0.00001405 0.00498708 + layer.1.k_cache 0.56163917 3.23252442 + layer.1.v_cache 0.00000554 0.00207947 + layer.2.k_cache 0.01059573 0.43538357 + layer.2.v_cache 0.00001807 0.00554156 + layer.3.k_cache 0.04944291 1.92786654 + layer.3.v_cache 0.00001950 0.00629090 + layer.4.k_cache 0.00071267 0.11908308 + layer.4.v_cache 0.00005817 0.01276438 + layer.4.output 0.00956290 197.62429425 + ------------------------------------------------------------------------------------- + TOTAL 0.04975818 83.03443544 + (elements=2,367,488) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2367488 +Total Bytes 458428 +BPFP 1.5491 bits/point +EBPFP 3.0982 equivalent bits/point +MSE 83.034435 +---------------------- -------------------------------------------------------- +Time: 0.853s Load: 0.010s, Pack+Encode: 0.342s, Decode+Unpack: 0.502s +---------------------- -------------------------------------------------------- +💾 Converting with 83.0344 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-20.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-20.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-204.zst (26/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-204.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 312, 128) +Output shape: (1, 312, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.0.v_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.1.k_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.1.v_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.2.k_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.2.v_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.3.k_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.3.v_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.4.k_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.4.v_cache: torch.Size([1, 4, 312, 128]) -> torch.Size([1, 1, 312, 512]) + layer.4.output: torch.Size([1, 312, 3584]) -> torch.Size([1, 1, 312, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 16,108B, BPFP=0.8067 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 51,064B, BPFP=2.5573 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,816B, BPFP=1.3930 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,836B, BPFP=2.4958 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,504B, BPFP=1.6278 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 49,268B, BPFP=2.4673 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,744B, BPFP=1.5397 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,916B, BPFP=2.4998 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,968B, BPFP=2.1518 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 48,184B, BPFP=2.4131 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 112,776B, BPFP=0.8068 +⌛️ [2/4] FRONTEND: Frontend time: 0.339s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 312, 128]) + layer.0.v_cache: torch.Size([1, 4, 312, 128]) + layer.1.k_cache: torch.Size([1, 4, 312, 128]) + layer.1.v_cache: torch.Size([1, 4, 312, 128]) + layer.2.k_cache: torch.Size([1, 4, 312, 128]) + layer.2.v_cache: torch.Size([1, 4, 312, 128]) + layer.3.k_cache: torch.Size([1, 4, 312, 128]) + layer.3.v_cache: torch.Size([1, 4, 312, 128]) + layer.4.k_cache: torch.Size([1, 4, 312, 128]) + layer.4.v_cache: torch.Size([1, 4, 312, 128]) + layer.4.output: torch.Size([1, 312, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.482s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 312, 128]) + layer.0.v_cache: torch.Size([1, 4, 312, 128]) + layer.1.k_cache: torch.Size([1, 4, 312, 128]) + layer.1.v_cache: torch.Size([1, 4, 312, 128]) + layer.2.k_cache: torch.Size([1, 4, 312, 128]) + layer.2.v_cache: torch.Size([1, 4, 312, 128]) + layer.3.k_cache: torch.Size([1, 4, 312, 128]) + layer.3.v_cache: torch.Size([1, 4, 312, 128]) + layer.4.k_cache: torch.Size([1, 4, 312, 128]) + layer.4.v_cache: torch.Size([1, 4, 312, 128]) + layer.4.output: torch.Size([1, 312, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14984696 21.77696658 + layer.0.v_cache 0.00001474 0.00498181 + layer.1.k_cache 0.67859346 3.00555576 + layer.1.v_cache 0.00000581 0.00210774 + layer.2.k_cache 0.01626646 0.43641521 + layer.2.v_cache 0.00001951 0.00557147 + layer.3.k_cache 0.02914289 1.82103964 + layer.3.v_cache 0.00001919 0.00616427 + layer.4.k_cache 0.00071665 0.11702140 + layer.4.v_cache 0.00005096 0.01191704 + layer.4.output 0.04642455 170.04897837 + ------------------------------------------------------------------------------------- + TOTAL 0.07056756 71.61944644 + (elements=2,715,648) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2715648 +Total Bytes 511184 +BPFP 1.5059 bits/point +EBPFP 3.0118 equivalent bits/point +MSE 71.619446 +---------------------- -------------------------------------------------------- +Time: 0.832s Load: 0.010s, Pack+Encode: 0.339s, Decode+Unpack: 0.482s +---------------------- -------------------------------------------------------- +💾 Converting with 71.6194 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-204.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-204.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-207.zst (27/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-207.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 269, 128) +Output shape: (1, 269, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.output: torch.Size([1, 269, 3584]) -> torch.Size([1, 1, 269, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,360B, BPFP=0.8341 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,560B, BPFP=2.7045 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,080B, BPFP=1.4568 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,392B, BPFP=2.6366 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,116B, BPFP=1.6912 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,776B, BPFP=2.6008 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,984B, BPFP=1.5674 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,272B, BPFP=2.6296 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,932B, BPFP=2.2614 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,964B, BPFP=2.5537 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 93,952B, BPFP=0.7796 +⌛️ [2/4] FRONTEND: Frontend time: 0.323s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.484s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15823836 22.98374877 + layer.0.v_cache 0.00001502 0.00492249 + layer.1.k_cache 0.53668020 3.09286919 + layer.1.v_cache 0.00000576 0.00206437 + layer.2.k_cache 0.01018995 0.43858630 + layer.2.v_cache 0.00002218 0.00567996 + layer.3.k_cache 0.05410197 2.06002978 + layer.3.v_cache 0.00002001 0.00631207 + layer.4.k_cache 0.00069143 0.11525985 + layer.4.v_cache 0.00005304 0.01264009 + layer.4.output 0.01041225 204.29789565 + ------------------------------------------------------------------------------------- + TOTAL 0.04899433 85.81219896 + (elements=2,341,376) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2341376 +Total Bytes 454388 +BPFP 1.5526 bits/point +EBPFP 3.1051 equivalent bits/point +MSE 85.812199 +---------------------- -------------------------------------------------------- +Time: 0.815s Load: 0.008s, Pack+Encode: 0.323s, Decode+Unpack: 0.484s +---------------------- -------------------------------------------------------- +💾 Converting with 85.8122 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-207.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-207.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-211.zst (28/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-211.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 329, 128) +Output shape: (1, 329, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.0.v_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.1.k_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.1.v_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.2.k_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.2.v_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.3.k_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.3.v_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.4.k_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.4.v_cache: torch.Size([1, 4, 329, 128]) -> torch.Size([1, 1, 329, 512]) + layer.4.output: torch.Size([1, 329, 3584]) -> torch.Size([1, 1, 329, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 17,156B, BPFP=0.8148 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 56,632B, BPFP=2.6896 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 30,032B, BPFP=1.4263 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 55,304B, BPFP=2.6265 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 35,088B, BPFP=1.6664 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 54,336B, BPFP=2.5805 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 32,860B, BPFP=1.5606 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 55,064B, BPFP=2.6151 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 47,096B, BPFP=2.2367 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 53,436B, BPFP=2.5378 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 127,220B, BPFP=0.8631 +⌛️ [2/4] FRONTEND: Frontend time: 0.640s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 329, 128]) + layer.0.v_cache: torch.Size([1, 4, 329, 128]) + layer.1.k_cache: torch.Size([1, 4, 329, 128]) + layer.1.v_cache: torch.Size([1, 4, 329, 128]) + layer.2.k_cache: torch.Size([1, 4, 329, 128]) + layer.2.v_cache: torch.Size([1, 4, 329, 128]) + layer.3.k_cache: torch.Size([1, 4, 329, 128]) + layer.3.v_cache: torch.Size([1, 4, 329, 128]) + layer.4.k_cache: torch.Size([1, 4, 329, 128]) + layer.4.v_cache: torch.Size([1, 4, 329, 128]) + layer.4.output: torch.Size([1, 329, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.598s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 329, 128]) + layer.0.v_cache: torch.Size([1, 4, 329, 128]) + layer.1.k_cache: torch.Size([1, 4, 329, 128]) + layer.1.v_cache: torch.Size([1, 4, 329, 128]) + layer.2.k_cache: torch.Size([1, 4, 329, 128]) + layer.2.v_cache: torch.Size([1, 4, 329, 128]) + layer.3.k_cache: torch.Size([1, 4, 329, 128]) + layer.3.v_cache: torch.Size([1, 4, 329, 128]) + layer.4.k_cache: torch.Size([1, 4, 329, 128]) + layer.4.v_cache: torch.Size([1, 4, 329, 128]) + layer.4.output: torch.Size([1, 329, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11391156 22.37612052 + layer.0.v_cache 0.00001522 0.00506386 + layer.1.k_cache 0.63933222 2.94981500 + layer.1.v_cache 0.00000652 0.00216090 + layer.2.k_cache 0.01643764 0.45452134 + layer.2.v_cache 0.00001873 0.00542423 + layer.3.k_cache 0.02826017 1.97969589 + layer.3.v_cache 0.00001876 0.00618190 + layer.4.k_cache 0.00074607 0.11702371 + layer.4.v_cache 0.00005483 0.01257552 + layer.4.output 3.44632065 160.88922058 + ------------------------------------------------------------------------------------- + TOTAL 1.46606155 67.89018394 + (elements=2,863,616) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2863616 +Total Bytes 564224 +BPFP 1.5763 bits/point +EBPFP 3.1525 equivalent bits/point +MSE 67.890184 +---------------------- -------------------------------------------------------- +Time: 1.249s Load: 0.011s, Pack+Encode: 0.640s, Decode+Unpack: 0.598s +---------------------- -------------------------------------------------------- +💾 Converting with 67.8902 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-211.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-211.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-215.zst (29/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-215.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.012s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 279, 128) +Output shape: (1, 279, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.output: torch.Size([1, 279, 3584]) -> torch.Size([1, 1, 279, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,668B, BPFP=0.8215 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,408B, BPFP=2.6550 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,436B, BPFP=1.4245 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,064B, BPFP=2.5797 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,872B, BPFP=1.6729 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,600B, BPFP=2.5538 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,044B, BPFP=1.5706 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,212B, BPFP=2.5880 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,832B, BPFP=2.2307 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,588B, BPFP=2.4971 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 94,892B, BPFP=0.7592 +⌛️ [2/4] FRONTEND: Frontend time: 0.399s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.531s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13536828 22.69965278 + layer.0.v_cache 0.00001433 0.00502630 + layer.1.k_cache 0.57185238 3.10704341 + layer.1.v_cache 0.00000558 0.00204407 + layer.2.k_cache 0.00766427 0.45136067 + layer.2.v_cache 0.00001891 0.00573222 + layer.3.k_cache 0.04143807 1.93773167 + layer.3.v_cache 0.00001947 0.00652933 + layer.4.k_cache 0.00070404 0.11392101 + layer.4.v_cache 0.00005498 0.01263033 + layer.4.output 0.00948056 194.92209101 + ------------------------------------------------------------------------------------- + TOTAL 0.04844142 81.92919464 + (elements=2,428,416) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2428416 +Total Bytes 462616 +BPFP 1.5240 bits/point +EBPFP 3.0480 equivalent bits/point +MSE 81.929195 +---------------------- -------------------------------------------------------- +Time: 0.942s Load: 0.012s, Pack+Encode: 0.399s, Decode+Unpack: 0.531s +---------------------- -------------------------------------------------------- +💾 Converting with 81.9292 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-215.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-215.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-216.zst (30/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-216.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 276, 128) +Output shape: (1, 276, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.output: torch.Size([1, 276, 3584]) -> torch.Size([1, 1, 276, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,340B, BPFP=0.8118 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,084B, BPFP=2.6655 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,144B, BPFP=1.4235 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,832B, BPFP=2.5947 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,648B, BPFP=1.6784 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,312B, BPFP=2.5652 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,680B, BPFP=1.5670 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,820B, BPFP=2.5940 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,596B, BPFP=2.2416 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,432B, BPFP=2.5154 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,240B, BPFP=0.8188 +⌛️ [2/4] FRONTEND: Frontend time: 0.376s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.544s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15132187 21.84736434 + layer.0.v_cache 0.00001456 0.00487697 + layer.1.k_cache 0.61798344 2.99995157 + layer.1.v_cache 0.00000569 0.00203435 + layer.2.k_cache 0.01686249 0.43409552 + layer.2.v_cache 0.00001898 0.00578150 + layer.3.k_cache 0.04401700 1.98066822 + layer.3.v_cache 0.00001935 0.00627686 + layer.4.k_cache 0.00071189 0.11655573 + layer.4.v_cache 0.00005111 0.01224034 + layer.4.output 0.01058244 193.43651333 + ------------------------------------------------------------------------------------- + TOTAL 0.05324020 81.26267286 + (elements=2,402,304) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2402304 +Total Bytes 466128 +BPFP 1.5523 bits/point +EBPFP 3.1045 equivalent bits/point +MSE 81.262673 +---------------------- -------------------------------------------------------- +Time: 0.929s Load: 0.009s, Pack+Encode: 0.376s, Decode+Unpack: 0.544s +---------------------- -------------------------------------------------------- +💾 Converting with 81.2627 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-216.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-216.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-22.zst (31/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-22.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 276, 128) +Output shape: (1, 276, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.output: torch.Size([1, 276, 3584]) -> torch.Size([1, 1, 276, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,588B, BPFP=0.8259 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,236B, BPFP=2.6741 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,428B, BPFP=1.4395 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,020B, BPFP=2.6053 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,920B, BPFP=1.6938 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,516B, BPFP=2.5768 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,756B, BPFP=1.5713 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,152B, BPFP=2.6128 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,700B, BPFP=2.2475 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,624B, BPFP=2.5263 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,960B, BPFP=0.8003 +⌛️ [2/4] FRONTEND: Frontend time: 0.379s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.532s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12581717 21.87652499 + layer.0.v_cache 0.00001375 0.00508079 + layer.1.k_cache 0.57814745 3.09639486 + layer.1.v_cache 0.00000592 0.00213075 + layer.2.k_cache 0.00733882 0.43898336 + layer.2.v_cache 0.00002058 0.00581122 + layer.3.k_cache 0.02353939 1.80282018 + layer.3.v_cache 0.00002077 0.00648356 + layer.4.k_cache 0.00069182 0.11970588 + layer.4.v_cache 0.00005027 0.01249324 + layer.4.output 0.01015823 189.69526398 + ------------------------------------------------------------------------------------- + TOTAL 0.04745609 79.71960451 + (elements=2,402,304) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2402304 +Total Bytes 465900 +BPFP 1.5515 bits/point +EBPFP 3.1030 equivalent bits/point +MSE 79.719605 +---------------------- -------------------------------------------------------- +Time: 0.921s Load: 0.009s, Pack+Encode: 0.379s, Decode+Unpack: 0.532s +---------------------- -------------------------------------------------------- +💾 Converting with 79.7196 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-22.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-22.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-221.zst (32/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-221.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 274, 128) +Output shape: (1, 274, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.output: torch.Size([1, 274, 3584]) -> torch.Size([1, 1, 274, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,364B, BPFP=0.8191 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,880B, BPFP=2.6734 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,152B, BPFP=1.4343 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,792B, BPFP=2.6113 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,648B, BPFP=1.6907 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,280B, BPFP=2.5821 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,632B, BPFP=1.5757 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,992B, BPFP=2.6227 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,612B, BPFP=2.2589 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,556B, BPFP=2.5408 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,544B, BPFP=0.8109 +⌛️ [2/4] FRONTEND: Frontend time: 0.341s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.output: torch.Size([1, 274, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.486s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.output: torch.Size([1, 274, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13625057 22.68231959 + layer.0.v_cache 0.00001362 0.00485116 + layer.1.k_cache 0.57512581 2.96127698 + layer.1.v_cache 0.00000584 0.00212412 + layer.2.k_cache 0.00872732 0.43227219 + layer.2.v_cache 0.00001851 0.00554292 + layer.3.k_cache 0.04362779 1.87204759 + layer.3.v_cache 0.00001925 0.00641014 + layer.4.k_cache 0.00073619 0.11510498 + layer.4.v_cache 0.00005148 0.01225443 + layer.4.output 0.00783937 199.11062956 + ------------------------------------------------------------------------------------- + TOTAL 0.04820306 83.63933006 + (elements=2,384,896) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2384896 +Total Bytes 464452 +BPFP 1.5580 bits/point +EBPFP 3.1160 equivalent bits/point +MSE 83.639330 +---------------------- -------------------------------------------------------- +Time: 0.837s Load: 0.010s, Pack+Encode: 0.341s, Decode+Unpack: 0.486s +---------------------- -------------------------------------------------------- +💾 Converting with 83.6393 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-221.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-221.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-222.zst (33/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-222.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 304, 128) +Output shape: (1, 304, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.0.v_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.1.k_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.1.v_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.2.k_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.2.v_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.3.k_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.3.v_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.4.k_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.4.v_cache: torch.Size([1, 4, 304, 128]) -> torch.Size([1, 1, 304, 512]) + layer.4.output: torch.Size([1, 304, 3584]) -> torch.Size([1, 1, 304, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 16,012B, BPFP=0.8230 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,528B, BPFP=2.5970 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,364B, BPFP=1.4065 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,276B, BPFP=2.5327 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,004B, BPFP=1.6449 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,668B, BPFP=2.5014 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,284B, BPFP=1.5565 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,444B, BPFP=2.5413 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,340B, BPFP=2.1762 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,788B, BPFP=2.4562 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 106,440B, BPFP=0.7815 +⌛️ [2/4] FRONTEND: Frontend time: 0.348s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 304, 128]) + layer.0.v_cache: torch.Size([1, 4, 304, 128]) + layer.1.k_cache: torch.Size([1, 4, 304, 128]) + layer.1.v_cache: torch.Size([1, 4, 304, 128]) + layer.2.k_cache: torch.Size([1, 4, 304, 128]) + layer.2.v_cache: torch.Size([1, 4, 304, 128]) + layer.3.k_cache: torch.Size([1, 4, 304, 128]) + layer.3.v_cache: torch.Size([1, 4, 304, 128]) + layer.4.k_cache: torch.Size([1, 4, 304, 128]) + layer.4.v_cache: torch.Size([1, 4, 304, 128]) + layer.4.output: torch.Size([1, 304, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.488s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 304, 128]) + layer.0.v_cache: torch.Size([1, 4, 304, 128]) + layer.1.k_cache: torch.Size([1, 4, 304, 128]) + layer.1.v_cache: torch.Size([1, 4, 304, 128]) + layer.2.k_cache: torch.Size([1, 4, 304, 128]) + layer.2.v_cache: torch.Size([1, 4, 304, 128]) + layer.3.k_cache: torch.Size([1, 4, 304, 128]) + layer.3.v_cache: torch.Size([1, 4, 304, 128]) + layer.4.k_cache: torch.Size([1, 4, 304, 128]) + layer.4.v_cache: torch.Size([1, 4, 304, 128]) + layer.4.output: torch.Size([1, 304, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16730670 22.44815063 + layer.0.v_cache 0.00001388 0.00482711 + layer.1.k_cache 0.64158324 3.10828119 + layer.1.v_cache 0.00000571 0.00196760 + layer.2.k_cache 0.02207795 0.45685914 + layer.2.v_cache 0.00001922 0.00526709 + layer.3.k_cache 0.04503629 2.01229577 + layer.3.v_cache 0.00001955 0.00607760 + layer.4.k_cache 0.00070478 0.11142569 + layer.4.v_cache 0.00005260 0.01155129 + layer.4.output 0.04763387 170.96719337 + ------------------------------------------------------------------------------------- + TOTAL 0.07119159 72.05512098 + (elements=2,646,016) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2646016 +Total Bytes 500148 +BPFP 1.5122 bits/point +EBPFP 3.0243 equivalent bits/point +MSE 72.055121 +---------------------- -------------------------------------------------------- +Time: 0.846s Load: 0.010s, Pack+Encode: 0.348s, Decode+Unpack: 0.488s +---------------------- -------------------------------------------------------- +💾 Converting with 72.0551 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-222.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-222.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-223.zst (34/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-223.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 277, 128) +Output shape: (1, 277, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.output: torch.Size([1, 277, 3584]) -> torch.Size([1, 1, 277, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,452B, BPFP=0.8152 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,000B, BPFP=2.6512 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,156B, BPFP=1.4190 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,800B, BPFP=2.5835 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,412B, BPFP=1.6591 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,280B, BPFP=2.5542 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,616B, BPFP=1.5578 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,960B, BPFP=2.5925 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,288B, BPFP=2.2162 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,364B, BPFP=2.5025 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,932B, BPFP=0.7972 +⌛️ [2/4] FRONTEND: Frontend time: 0.334s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.531s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15225300 22.83976443 + layer.0.v_cache 0.00001427 0.00465022 + layer.1.k_cache 0.52731114 3.31269126 + layer.1.v_cache 0.00000556 0.00200395 + layer.2.k_cache 0.01483519 0.43577733 + layer.2.v_cache 0.00001864 0.00547828 + layer.3.k_cache 0.04699480 1.89473397 + layer.3.v_cache 0.00001968 0.00598531 + layer.4.k_cache 0.00070142 0.11296679 + layer.4.v_cache 0.00005177 0.01181711 + layer.4.output 0.01077793 198.14991619 + ------------------------------------------------------------------------------------- + TOTAL 0.04809712 83.27501659 + (elements=2,411,008) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2411008 +Total Bytes 463260 +BPFP 1.5371 bits/point +EBPFP 3.0743 equivalent bits/point +MSE 83.275017 +---------------------- -------------------------------------------------------- +Time: 0.874s Load: 0.009s, Pack+Encode: 0.334s, Decode+Unpack: 0.531s +---------------------- -------------------------------------------------------- +💾 Converting with 83.2750 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-223.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-223.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-23.zst (35/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-23.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 277, 128) +Output shape: (1, 277, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.output: torch.Size([1, 277, 3584]) -> torch.Size([1, 1, 277, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,524B, BPFP=0.8193 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,068B, BPFP=2.6550 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,284B, BPFP=1.4262 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,996B, BPFP=2.5945 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,640B, BPFP=1.6719 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,548B, BPFP=2.5693 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,740B, BPFP=1.5648 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,092B, BPFP=2.6000 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,416B, BPFP=2.2234 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,528B, BPFP=2.5117 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,352B, BPFP=0.8087 +⌛️ [2/4] FRONTEND: Frontend time: 0.356s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.518s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15700974 22.44193860 + layer.0.v_cache 0.00001374 0.00473577 + layer.1.k_cache 0.52255304 3.22736433 + layer.1.v_cache 0.00000566 0.00196608 + layer.2.k_cache 0.01360668 0.44971805 + layer.2.v_cache 0.00001926 0.00547201 + layer.3.k_cache 0.08085050 1.93163578 + layer.3.v_cache 0.00001943 0.00610648 + layer.4.k_cache 0.00071029 0.11237822 + layer.4.v_cache 0.00005450 0.01190718 + layer.4.output 0.01014708 198.56406331 + ------------------------------------------------------------------------------------- + TOTAL 0.04975720 83.42009798 + (elements=2,411,008) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2411008 +Total Bytes 466188 +BPFP 1.5469 bits/point +EBPFP 3.0937 equivalent bits/point +MSE 83.420098 +---------------------- -------------------------------------------------------- +Time: 0.883s Load: 0.009s, Pack+Encode: 0.356s, Decode+Unpack: 0.518s +---------------------- -------------------------------------------------------- +💾 Converting with 83.4201 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-23.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-23.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-232.zst (36/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-232.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 285, 128) +Output shape: (1, 285, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.output: torch.Size([1, 285, 3584]) -> torch.Size([1, 1, 285, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,196B, BPFP=0.8331 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,032B, BPFP=2.6333 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,316B, BPFP=1.4428 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,832B, BPFP=2.5675 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,536B, BPFP=1.6741 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,224B, BPFP=2.5342 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,712B, BPFP=1.5741 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,892B, BPFP=2.5708 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,504B, BPFP=2.2206 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,280B, BPFP=2.4825 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,020B, BPFP=0.7912 +⌛️ [2/4] FRONTEND: Frontend time: 0.345s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.521s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16436556 22.68715221 + layer.0.v_cache 0.00001441 0.00495806 + layer.1.k_cache 0.62741763 2.99666684 + layer.1.v_cache 0.00000560 0.00207584 + layer.2.k_cache 0.01487184 0.47102998 + layer.2.v_cache 0.00002167 0.00553258 + layer.3.k_cache 0.02142572 2.09633125 + layer.3.v_cache 0.00001943 0.00618360 + layer.4.k_cache 0.00070776 0.11360359 + layer.4.v_cache 0.00005251 0.01186598 + layer.4.output 0.00894435 192.81959586 + ------------------------------------------------------------------------------------- + TOTAL 0.05244192 81.06662182 + (elements=2,480,640) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2480640 +Total Bytes 475544 +BPFP 1.5336 bits/point +EBPFP 3.0672 equivalent bits/point +MSE 81.066622 +---------------------- -------------------------------------------------------- +Time: 0.876s Load: 0.009s, Pack+Encode: 0.345s, Decode+Unpack: 0.521s +---------------------- -------------------------------------------------------- +💾 Converting with 81.0666 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-232.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-232.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-236.zst (37/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-236.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 281, 128) +Output shape: (1, 281, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.output: torch.Size([1, 281, 3584]) -> torch.Size([1, 1, 281, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,736B, BPFP=0.8194 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,356B, BPFP=2.6332 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,760B, BPFP=1.4324 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,072B, BPFP=2.5618 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,708B, BPFP=1.6519 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,512B, BPFP=2.5307 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,964B, BPFP=1.5549 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,092B, BPFP=2.5629 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,676B, BPFP=2.2062 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,632B, BPFP=2.4818 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,304B, BPFP=0.7809 +⌛️ [2/4] FRONTEND: Frontend time: 0.346s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.output: torch.Size([1, 281, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.476s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.output: torch.Size([1, 281, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11705062 22.74105976 + layer.0.v_cache 0.00001445 0.00494022 + layer.1.k_cache 0.56620099 3.17201249 + layer.1.v_cache 0.00000566 0.00203923 + layer.2.k_cache 0.01299965 0.42593959 + layer.2.v_cache 0.00001972 0.00539067 + layer.3.k_cache 0.05069733 1.93524974 + layer.3.v_cache 0.00001944 0.00636526 + layer.4.k_cache 0.00068883 0.11406026 + layer.4.v_cache 0.00005144 0.01171722 + layer.4.output 0.00838925 194.86885168 + ------------------------------------------------------------------------------------- + TOTAL 0.04743958 81.91180801 + (elements=2,445,824) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2445824 +Total Bytes 465812 +BPFP 1.5236 bits/point +EBPFP 3.0472 equivalent bits/point +MSE 81.911808 +---------------------- -------------------------------------------------------- +Time: 0.831s Load: 0.010s, Pack+Encode: 0.346s, Decode+Unpack: 0.476s +---------------------- -------------------------------------------------------- +💾 Converting with 81.9118 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-236.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-236.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-24.zst (38/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-24.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,944B, BPFP=0.8251 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,744B, BPFP=2.6360 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,240B, BPFP=1.4488 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,536B, BPFP=2.5693 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,392B, BPFP=1.6780 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,812B, BPFP=2.5294 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,564B, BPFP=1.5771 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,464B, BPFP=2.5654 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,392B, BPFP=2.2301 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,112B, BPFP=2.4907 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,156B, BPFP=0.8057 +⌛️ [2/4] FRONTEND: Frontend time: 0.328s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.470s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14370025 22.62943594 + layer.0.v_cache 0.00001382 0.00503112 + layer.1.k_cache 0.61448195 2.99410525 + layer.1.v_cache 0.00000570 0.00210881 + layer.2.k_cache 0.01208573 0.44141186 + layer.2.v_cache 0.00001823 0.00576718 + layer.3.k_cache 0.01319265 1.82915203 + layer.3.v_cache 0.00002242 0.00649635 + layer.4.k_cache 0.00071920 0.11654334 + layer.4.v_cache 0.00005145 0.01273605 + layer.4.output 0.00992508 188.64839412 + ------------------------------------------------------------------------------------- + TOTAL 0.05022159 79.32832628 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 474356 +BPFP 1.5406 bits/point +EBPFP 3.0812 equivalent bits/point +MSE 79.328326 +---------------------- -------------------------------------------------------- +Time: 0.808s Load: 0.009s, Pack+Encode: 0.328s, Decode+Unpack: 0.470s +---------------------- -------------------------------------------------------- +💾 Converting with 79.3283 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-24.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-24.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-245.zst (39/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-245.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 259, 128) +Output shape: (1, 259, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.0.v_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.1.k_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.1.v_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.2.k_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.2.v_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.3.k_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.3.v_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.4.k_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.4.v_cache: torch.Size([1, 4, 259, 128]) -> torch.Size([1, 1, 259, 512]) + layer.4.output: torch.Size([1, 259, 3584]) -> torch.Size([1, 1, 259, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,300B, BPFP=0.8024 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 45,696B, BPFP=2.7568 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 23,980B, BPFP=1.4467 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 44,444B, BPFP=2.6812 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 27,852B, BPFP=1.6803 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 43,648B, BPFP=2.6332 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,036B, BPFP=1.5707 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 44,224B, BPFP=2.6680 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,108B, BPFP=2.2990 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 42,968B, BPFP=2.5922 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 94,160B, BPFP=0.8115 +⌛️ [2/4] FRONTEND: Frontend time: 0.407s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 259, 128]) + layer.0.v_cache: torch.Size([1, 4, 259, 128]) + layer.1.k_cache: torch.Size([1, 4, 259, 128]) + layer.1.v_cache: torch.Size([1, 4, 259, 128]) + layer.2.k_cache: torch.Size([1, 4, 259, 128]) + layer.2.v_cache: torch.Size([1, 4, 259, 128]) + layer.3.k_cache: torch.Size([1, 4, 259, 128]) + layer.3.v_cache: torch.Size([1, 4, 259, 128]) + layer.4.k_cache: torch.Size([1, 4, 259, 128]) + layer.4.v_cache: torch.Size([1, 4, 259, 128]) + layer.4.output: torch.Size([1, 259, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.544s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 259, 128]) + layer.0.v_cache: torch.Size([1, 4, 259, 128]) + layer.1.k_cache: torch.Size([1, 4, 259, 128]) + layer.1.v_cache: torch.Size([1, 4, 259, 128]) + layer.2.k_cache: torch.Size([1, 4, 259, 128]) + layer.2.v_cache: torch.Size([1, 4, 259, 128]) + layer.3.k_cache: torch.Size([1, 4, 259, 128]) + layer.3.v_cache: torch.Size([1, 4, 259, 128]) + layer.4.k_cache: torch.Size([1, 4, 259, 128]) + layer.4.v_cache: torch.Size([1, 4, 259, 128]) + layer.4.output: torch.Size([1, 259, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13349269 22.46107701 + layer.0.v_cache 0.00001436 0.00494696 + layer.1.k_cache 0.50562990 2.90489615 + layer.1.v_cache 0.00000590 0.00207607 + layer.2.k_cache 0.00820730 0.42872785 + layer.2.v_cache 0.00001794 0.00541953 + layer.3.k_cache 0.03685973 1.91724705 + layer.3.v_cache 0.00002061 0.00614055 + layer.4.k_cache 0.00069539 0.11537967 + layer.4.v_cache 0.00005422 0.01216871 + layer.4.output 0.01018517 206.09278475 + ------------------------------------------------------------------------------------- + TOTAL 0.04448790 86.50044546 + (elements=2,254,336) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2254336 +Total Bytes 444416 +BPFP 1.5771 bits/point +EBPFP 3.1542 equivalent bits/point +MSE 86.500445 +---------------------- -------------------------------------------------------- +Time: 0.960s Load: 0.009s, Pack+Encode: 0.407s, Decode+Unpack: 0.544s +---------------------- -------------------------------------------------------- +💾 Converting with 86.5004 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-245.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-245.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-246.zst (40/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-246.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 270, 128) +Output shape: (1, 270, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.output: torch.Size([1, 270, 3584]) -> torch.Size([1, 1, 270, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,976B, BPFP=0.8088 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,772B, BPFP=2.7067 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,968B, BPFP=1.4449 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,488B, BPFP=2.6324 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,128B, BPFP=1.6856 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,048B, BPFP=2.6069 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,368B, BPFP=1.5838 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,492B, BPFP=2.6326 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,228B, BPFP=2.2701 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,104B, BPFP=2.5523 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 96,184B, BPFP=0.7952 +⌛️ [2/4] FRONTEND: Frontend time: 0.376s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.output: torch.Size([1, 270, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.525s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.output: torch.Size([1, 270, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12649573 22.54120913 + layer.0.v_cache 0.00001373 0.00503413 + layer.1.k_cache 0.51574391 3.02305773 + layer.1.v_cache 0.00000589 0.00213446 + layer.2.k_cache 0.01194659 0.43724752 + layer.2.v_cache 0.00001984 0.00584556 + layer.3.k_cache 0.04962628 1.97180764 + layer.3.v_cache 0.00002059 0.00679245 + layer.4.k_cache 0.00068541 0.11679298 + layer.4.v_cache 0.00005704 0.01309309 + layer.4.output 0.00772497 203.26430225 + ------------------------------------------------------------------------------------- + TOTAL 0.04462881 85.35136061 + (elements=2,350,080) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2350080 +Total Bytes 457756 +BPFP 1.5583 bits/point +EBPFP 3.1165 equivalent bits/point +MSE 85.351361 +---------------------- -------------------------------------------------------- +Time: 0.911s Load: 0.010s, Pack+Encode: 0.376s, Decode+Unpack: 0.525s +---------------------- -------------------------------------------------------- +💾 Converting with 85.3514 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-246.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-246.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-249.zst (41/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-249.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 269, 128) +Output shape: (1, 269, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) -> torch.Size([1, 1, 269, 512]) + layer.4.output: torch.Size([1, 269, 3584]) -> torch.Size([1, 1, 269, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,320B, BPFP=0.8318 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,540B, BPFP=2.7033 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,000B, BPFP=1.4521 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,236B, BPFP=2.6276 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,036B, BPFP=1.6866 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,708B, BPFP=2.5969 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,192B, BPFP=1.5795 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,376B, BPFP=2.6357 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,092B, BPFP=2.2707 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,956B, BPFP=2.5532 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 93,176B, BPFP=0.7732 +⌛️ [2/4] FRONTEND: Frontend time: 0.381s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.516s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 269, 128]) + layer.0.v_cache: torch.Size([1, 4, 269, 128]) + layer.1.k_cache: torch.Size([1, 4, 269, 128]) + layer.1.v_cache: torch.Size([1, 4, 269, 128]) + layer.2.k_cache: torch.Size([1, 4, 269, 128]) + layer.2.v_cache: torch.Size([1, 4, 269, 128]) + layer.3.k_cache: torch.Size([1, 4, 269, 128]) + layer.3.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.k_cache: torch.Size([1, 4, 269, 128]) + layer.4.v_cache: torch.Size([1, 4, 269, 128]) + layer.4.output: torch.Size([1, 269, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12210708 22.52843285 + layer.0.v_cache 0.00001342 0.00502371 + layer.1.k_cache 0.54942129 3.24238195 + layer.1.v_cache 0.00000547 0.00207177 + layer.2.k_cache 0.01053270 0.44800701 + layer.2.v_cache 0.00001820 0.00578595 + layer.3.k_cache 0.02850239 1.90371012 + layer.3.v_cache 0.00002491 0.00674329 + layer.4.k_cache 0.00067987 0.11814690 + layer.4.v_cache 0.00005179 0.01281527 + layer.4.output 0.01126663 204.28712493 + ------------------------------------------------------------------------------------- + TOTAL 0.04648373 85.78135255 + (elements=2,341,376) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2341376 +Total Bytes 453632 +BPFP 1.5500 bits/point +EBPFP 3.0999 equivalent bits/point +MSE 85.781353 +---------------------- -------------------------------------------------------- +Time: 0.907s Load: 0.010s, Pack+Encode: 0.381s, Decode+Unpack: 0.516s +---------------------- -------------------------------------------------------- +💾 Converting with 85.7814 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-249.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-249.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-250.zst (42/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-250.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 303, 128) +Output shape: (1, 303, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.output: torch.Size([1, 303, 3584]) -> torch.Size([1, 1, 303, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,700B, BPFP=0.8096 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,464B, BPFP=2.6023 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,440B, BPFP=1.4150 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,184B, BPFP=2.5363 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,764B, BPFP=1.6380 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,520B, BPFP=2.5021 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,096B, BPFP=1.5520 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,992B, BPFP=2.5264 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,200B, BPFP=2.1762 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,304B, BPFP=2.4394 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,684B, BPFP=0.7196 +⌛️ [2/4] FRONTEND: Frontend time: 0.379s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.output: torch.Size([1, 303, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.502s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.output: torch.Size([1, 303, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14443027 22.00655070 + layer.0.v_cache 0.00001375 0.00507383 + layer.1.k_cache 0.62837083 3.04716406 + layer.1.v_cache 0.00000570 0.00213811 + layer.2.k_cache 0.01557094 0.45984415 + layer.2.v_cache 0.00001871 0.00561738 + layer.3.k_cache 0.04676386 1.89719504 + layer.3.v_cache 0.00001981 0.00635658 + layer.4.k_cache 0.00070582 0.11611512 + layer.4.v_cache 0.00005437 0.01217391 + layer.4.output 0.04838977 173.98165665 + ------------------------------------------------------------------------------------- + TOTAL 0.06909897 73.26057797 + (elements=2,637,312) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2637312 +Total Bytes 489348 +BPFP 1.4844 bits/point +EBPFP 2.9688 equivalent bits/point +MSE 73.260578 +---------------------- -------------------------------------------------------- +Time: 0.892s Load: 0.011s, Pack+Encode: 0.379s, Decode+Unpack: 0.502s +---------------------- -------------------------------------------------------- +💾 Converting with 73.2606 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-250.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-250.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-252.zst (43/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-252.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,984B, BPFP=0.8273 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,860B, BPFP=2.6424 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,088B, BPFP=1.4404 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,508B, BPFP=2.5678 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,312B, BPFP=1.6736 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,008B, BPFP=2.5402 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,588B, BPFP=1.5784 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,652B, BPFP=2.5758 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,392B, BPFP=2.2301 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,108B, BPFP=2.4905 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,844B, BPFP=0.7954 +⌛️ [2/4] FRONTEND: Frontend time: 0.358s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.504s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13293209 23.34198322 + layer.0.v_cache 0.00001388 0.00496032 + layer.1.k_cache 0.56267701 3.01694621 + layer.1.v_cache 0.00000566 0.00206846 + layer.2.k_cache 0.01145046 0.44678961 + layer.2.v_cache 0.00001921 0.00568211 + layer.3.k_cache 0.02674564 1.85844826 + layer.3.v_cache 0.00001926 0.00641234 + layer.4.k_cache 0.00069135 0.11676275 + layer.4.v_cache 0.00005493 0.01278402 + layer.4.output 0.00973303 186.18232585 + ------------------------------------------------------------------------------------- + TOTAL 0.04722004 78.35818342 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 473344 +BPFP 1.5373 bits/point +EBPFP 3.0746 equivalent bits/point +MSE 78.358183 +---------------------- -------------------------------------------------------- +Time: 0.871s Load: 0.009s, Pack+Encode: 0.358s, Decode+Unpack: 0.504s +---------------------- -------------------------------------------------------- +💾 Converting with 78.3582 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-252.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-252.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-253.zst (44/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-253.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 277, 128) +Output shape: (1, 277, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.output: torch.Size([1, 277, 3584]) -> torch.Size([1, 1, 277, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,464B, BPFP=0.8159 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,132B, BPFP=2.6586 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,344B, BPFP=1.4296 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,096B, BPFP=2.6002 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,736B, BPFP=1.6773 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,412B, BPFP=2.5616 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,908B, BPFP=1.5742 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,004B, BPFP=2.5950 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,872B, BPFP=2.2491 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,556B, BPFP=2.5133 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 96,268B, BPFP=0.7758 +⌛️ [2/4] FRONTEND: Frontend time: 0.375s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.571s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15182159 23.05684158 + layer.0.v_cache 0.00001420 0.00508952 + layer.1.k_cache 0.56224567 3.12344658 + layer.1.v_cache 0.00000588 0.00219124 + layer.2.k_cache 0.00652889 0.42668526 + layer.2.v_cache 0.00001844 0.00581146 + layer.3.k_cache 0.06011557 1.87574300 + layer.3.v_cache 0.00001921 0.00676954 + layer.4.k_cache 0.00068317 0.11546827 + layer.4.v_cache 0.00005096 0.01249949 + layer.4.output 0.01130176 198.20255286 + ------------------------------------------------------------------------------------- + TOTAL 0.05062446 83.29696565 + (elements=2,411,008) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2411008 +Total Bytes 462792 +BPFP 1.5356 bits/point +EBPFP 3.0712 equivalent bits/point +MSE 83.296966 +---------------------- -------------------------------------------------------- +Time: 0.957s Load: 0.011s, Pack+Encode: 0.375s, Decode+Unpack: 0.571s +---------------------- -------------------------------------------------------- +💾 Converting with 83.2970 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-253.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-253.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-254.zst (45/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-254.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 285, 128) +Output shape: (1, 285, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) -> torch.Size([1, 1, 285, 512]) + layer.4.output: torch.Size([1, 285, 3584]) -> torch.Size([1, 1, 285, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,432B, BPFP=0.8461 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,276B, BPFP=2.6467 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,356B, BPFP=1.4450 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,032B, BPFP=2.5785 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,824B, BPFP=1.6899 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,380B, BPFP=2.5428 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,088B, BPFP=1.5947 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,008B, BPFP=2.5772 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,812B, BPFP=2.2375 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,644B, BPFP=2.5024 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,184B, BPFP=0.8160 +⌛️ [2/4] FRONTEND: Frontend time: 0.333s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.473s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 285, 128]) + layer.0.v_cache: torch.Size([1, 4, 285, 128]) + layer.1.k_cache: torch.Size([1, 4, 285, 128]) + layer.1.v_cache: torch.Size([1, 4, 285, 128]) + layer.2.k_cache: torch.Size([1, 4, 285, 128]) + layer.2.v_cache: torch.Size([1, 4, 285, 128]) + layer.3.k_cache: torch.Size([1, 4, 285, 128]) + layer.3.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.k_cache: torch.Size([1, 4, 285, 128]) + layer.4.v_cache: torch.Size([1, 4, 285, 128]) + layer.4.output: torch.Size([1, 285, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13053865 22.99331654 + layer.0.v_cache 0.00001414 0.00541947 + layer.1.k_cache 0.56363723 3.11086940 + layer.1.v_cache 0.00000590 0.00225220 + layer.2.k_cache 0.01033439 0.45098834 + layer.2.v_cache 0.00002015 0.00599211 + layer.3.k_cache 0.03655195 1.87832695 + layer.3.v_cache 0.00002026 0.00668063 + layer.4.k_cache 0.00068516 0.12132531 + layer.4.v_cache 0.00005378 0.01331009 + layer.4.output 0.00863618 192.26904762 + ------------------------------------------------------------------------------------- + TOTAL 0.04719499 80.85128320 + (elements=2,480,640) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2480640 +Total Bytes 481036 +BPFP 1.5513 bits/point +EBPFP 3.1027 equivalent bits/point +MSE 80.851283 +---------------------- -------------------------------------------------------- +Time: 0.815s Load: 0.010s, Pack+Encode: 0.333s, Decode+Unpack: 0.473s +---------------------- -------------------------------------------------------- +💾 Converting with 80.8513 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-254.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-254.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-255.zst (46/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-255.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 277, 128) +Output shape: (1, 277, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.output: torch.Size([1, 277, 3584]) -> torch.Size([1, 1, 277, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,588B, BPFP=0.8229 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,168B, BPFP=2.6606 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,288B, BPFP=1.4264 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,952B, BPFP=2.5921 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,604B, BPFP=1.6699 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,292B, BPFP=2.5548 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,788B, BPFP=1.5675 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,084B, BPFP=2.5995 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,580B, BPFP=2.2326 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,496B, BPFP=2.5099 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,600B, BPFP=0.8107 +⌛️ [2/4] FRONTEND: Frontend time: 0.344s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.482s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15908699 22.37964308 + layer.0.v_cache 0.00001433 0.00500451 + layer.1.k_cache 0.53822481 3.06818090 + layer.1.v_cache 0.00000570 0.00208758 + layer.2.k_cache 0.01139215 0.43449961 + layer.2.v_cache 0.00001845 0.00558706 + layer.3.k_cache 0.02907990 1.98549038 + layer.3.v_cache 0.00001963 0.00637855 + layer.4.k_cache 0.00069750 0.11275861 + layer.4.v_cache 0.00005397 0.01227187 + layer.4.output 0.01069233 198.68809631 + ------------------------------------------------------------------------------------- + TOTAL 0.04784940 83.46050449 + (elements=2,411,008) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2411008 +Total Bytes 466440 +BPFP 1.5477 bits/point +EBPFP 3.0954 equivalent bits/point +MSE 83.460504 +---------------------- -------------------------------------------------------- +Time: 0.834s Load: 0.009s, Pack+Encode: 0.344s, Decode+Unpack: 0.482s +---------------------- -------------------------------------------------------- +💾 Converting with 83.4605 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-255.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-255.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-258.zst (47/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-258.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 286, 128) +Output shape: (1, 286, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.output: torch.Size([1, 286, 3584]) -> torch.Size([1, 1, 286, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,164B, BPFP=0.8285 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,296B, BPFP=2.6385 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,344B, BPFP=1.4392 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,036B, BPFP=2.5697 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,868B, BPFP=1.6864 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,500B, BPFP=2.5404 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,072B, BPFP=1.5883 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,128B, BPFP=2.5747 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,752B, BPFP=2.2264 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,572B, BPFP=2.4897 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,304B, BPFP=0.7985 +⌛️ [2/4] FRONTEND: Frontend time: 0.316s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.468s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16084659 23.16338300 + layer.0.v_cache 0.00001440 0.00490826 + layer.1.k_cache 0.59028764 3.11259962 + layer.1.v_cache 0.00000559 0.00202835 + layer.2.k_cache 0.00929020 0.43089572 + layer.2.v_cache 0.00002205 0.00565341 + layer.3.k_cache 0.04382113 1.94054301 + layer.3.v_cache 0.00002103 0.00640718 + layer.4.k_cache 0.00069271 0.11622145 + layer.4.v_cache 0.00005052 0.01221461 + layer.4.output 0.00984751 192.23033217 + ------------------------------------------------------------------------------------- + TOTAL 0.05141085 80.84748116 + (elements=2,489,344) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2489344 +Total Bytes 479036 +BPFP 1.5395 bits/point +EBPFP 3.0790 equivalent bits/point +MSE 80.847481 +---------------------- -------------------------------------------------------- +Time: 0.794s Load: 0.010s, Pack+Encode: 0.316s, Decode+Unpack: 0.468s +---------------------- -------------------------------------------------------- +💾 Converting with 80.8475 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-258.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-258.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-260.zst (48/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-260.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 288, 128) +Output shape: (1, 288, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.output: torch.Size([1, 288, 3584]) -> torch.Size([1, 1, 288, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,880B, BPFP=0.8073 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,368B, BPFP=2.6241 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,112B, BPFP=1.4167 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,156B, BPFP=2.5584 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,440B, BPFP=1.6515 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,396B, BPFP=2.5171 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,748B, BPFP=1.5597 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,156B, BPFP=2.5584 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,536B, BPFP=2.1992 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,540B, BPFP=2.4707 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 105,624B, BPFP=0.8186 +⌛️ [2/4] FRONTEND: Frontend time: 0.305s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.523s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14211286 22.15341356 + layer.0.v_cache 0.00001516 0.00479755 + layer.1.k_cache 0.60291523 3.05334918 + layer.1.v_cache 0.00000570 0.00203427 + layer.2.k_cache 0.01041983 0.42131318 + layer.2.v_cache 0.00001835 0.00525205 + layer.3.k_cache 0.03896446 1.76885880 + layer.3.v_cache 0.00001926 0.00618985 + layer.4.k_cache 0.00071470 0.11324518 + layer.4.v_cache 0.00005236 0.01224044 + layer.4.output 0.00754456 186.62262835 + ------------------------------------------------------------------------------------- + TOTAL 0.04988528 78.46465250 + (elements=2,506,752) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2506752 +Total Bytes 480956 +BPFP 1.5349 bits/point +EBPFP 3.0698 equivalent bits/point +MSE 78.464652 +---------------------- -------------------------------------------------------- +Time: 0.838s Load: 0.010s, Pack+Encode: 0.305s, Decode+Unpack: 0.523s +---------------------- -------------------------------------------------------- +💾 Converting with 78.4647 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-260.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-260.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-264.zst (49/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-264.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 296, 128) +Output shape: (1, 296, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.0.v_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.1.k_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.1.v_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.2.k_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.2.v_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.3.k_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.3.v_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.4.k_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.4.v_cache: torch.Size([1, 4, 296, 128]) -> torch.Size([1, 1, 296, 512]) + layer.4.output: torch.Size([1, 296, 3584]) -> torch.Size([1, 1, 296, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,628B, BPFP=0.8250 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,668B, BPFP=2.6218 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,004B, BPFP=1.4255 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,496B, BPFP=2.5600 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,432B, BPFP=1.6592 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,876B, BPFP=2.5272 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,680B, BPFP=1.5667 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,676B, BPFP=2.5695 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,820B, BPFP=2.2076 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,928B, BPFP=2.4772 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,480B, BPFP=0.7502 +⌛️ [2/4] FRONTEND: Frontend time: 0.364s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 296, 128]) + layer.0.v_cache: torch.Size([1, 4, 296, 128]) + layer.1.k_cache: torch.Size([1, 4, 296, 128]) + layer.1.v_cache: torch.Size([1, 4, 296, 128]) + layer.2.k_cache: torch.Size([1, 4, 296, 128]) + layer.2.v_cache: torch.Size([1, 4, 296, 128]) + layer.3.k_cache: torch.Size([1, 4, 296, 128]) + layer.3.v_cache: torch.Size([1, 4, 296, 128]) + layer.4.k_cache: torch.Size([1, 4, 296, 128]) + layer.4.v_cache: torch.Size([1, 4, 296, 128]) + layer.4.output: torch.Size([1, 296, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.503s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 296, 128]) + layer.0.v_cache: torch.Size([1, 4, 296, 128]) + layer.1.k_cache: torch.Size([1, 4, 296, 128]) + layer.1.v_cache: torch.Size([1, 4, 296, 128]) + layer.2.k_cache: torch.Size([1, 4, 296, 128]) + layer.2.v_cache: torch.Size([1, 4, 296, 128]) + layer.3.k_cache: torch.Size([1, 4, 296, 128]) + layer.3.v_cache: torch.Size([1, 4, 296, 128]) + layer.4.k_cache: torch.Size([1, 4, 296, 128]) + layer.4.v_cache: torch.Size([1, 4, 296, 128]) + layer.4.output: torch.Size([1, 296, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14616069 22.43389398 + layer.0.v_cache 0.00001366 0.00492925 + layer.1.k_cache 0.61022733 3.01680817 + layer.1.v_cache 0.00000576 0.00200705 + layer.2.k_cache 0.01378242 0.43701172 + layer.2.v_cache 0.00001907 0.00554590 + layer.3.k_cache 0.03184220 1.95629099 + layer.3.v_cache 0.00001934 0.00622418 + layer.4.k_cache 0.00072724 0.11418703 + layer.4.v_cache 0.00005080 0.01176282 + layer.4.output 0.04784788 180.10199988 + ------------------------------------------------------------------------------------- + TOTAL 0.06692845 75.80603884 + (elements=2,576,384) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2576384 +Total Bytes 486688 +BPFP 1.5112 bits/point +EBPFP 3.0225 equivalent bits/point +MSE 75.806039 +---------------------- -------------------------------------------------------- +Time: 0.877s Load: 0.009s, Pack+Encode: 0.364s, Decode+Unpack: 0.503s +---------------------- -------------------------------------------------------- +💾 Converting with 75.8060 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-264.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-264.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-265.zst (50/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-265.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 298, 128) +Output shape: (1, 298, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) -> torch.Size([1, 1, 298, 512]) + layer.4.output: torch.Size([1, 298, 3584]) -> torch.Size([1, 1, 298, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,720B, BPFP=0.8242 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,112B, BPFP=2.6275 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,300B, BPFP=1.4314 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,916B, BPFP=2.5648 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,732B, BPFP=1.6638 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,276B, BPFP=2.5312 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,296B, BPFP=1.5885 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,972B, BPFP=2.5677 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,092B, BPFP=2.2070 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,332B, BPFP=2.4818 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,964B, BPFP=0.7712 +⌛️ [2/4] FRONTEND: Frontend time: 0.331s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.497s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 298, 128]) + layer.0.v_cache: torch.Size([1, 4, 298, 128]) + layer.1.k_cache: torch.Size([1, 4, 298, 128]) + layer.1.v_cache: torch.Size([1, 4, 298, 128]) + layer.2.k_cache: torch.Size([1, 4, 298, 128]) + layer.2.v_cache: torch.Size([1, 4, 298, 128]) + layer.3.k_cache: torch.Size([1, 4, 298, 128]) + layer.3.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.k_cache: torch.Size([1, 4, 298, 128]) + layer.4.v_cache: torch.Size([1, 4, 298, 128]) + layer.4.output: torch.Size([1, 298, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15387250 22.93599092 + layer.0.v_cache 0.00001382 0.00490207 + layer.1.k_cache 0.63844893 3.19497414 + layer.1.v_cache 0.00000583 0.00212940 + layer.2.k_cache 0.01105453 0.45139354 + layer.2.v_cache 0.00001885 0.00573792 + layer.3.k_cache 0.03893430 1.93715017 + layer.3.v_cache 0.00001923 0.00651598 + layer.4.k_cache 0.00070257 0.11388000 + layer.4.v_cache 0.00005146 0.01208425 + layer.4.output 0.04873301 180.30148909 + ------------------------------------------------------------------------------------- + TOTAL 0.06966195 75.92795189 + (elements=2,593,792) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2593792 +Total Bytes 493712 +BPFP 1.5227 bits/point +EBPFP 3.0455 equivalent bits/point +MSE 75.927952 +---------------------- -------------------------------------------------------- +Time: 0.837s Load: 0.010s, Pack+Encode: 0.331s, Decode+Unpack: 0.497s +---------------------- -------------------------------------------------------- +💾 Converting with 75.9280 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-265.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-265.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-266.zst (51/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-266.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 271, 128) +Output shape: (1, 271, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) -> torch.Size([1, 1, 271, 512]) + layer.4.output: torch.Size([1, 271, 3584]) -> torch.Size([1, 1, 271, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,192B, BPFP=0.8183 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,872B, BPFP=2.7025 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,008B, BPFP=1.4419 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,736B, BPFP=2.6370 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,340B, BPFP=1.6917 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,072B, BPFP=2.5987 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,528B, BPFP=1.5872 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,672B, BPFP=2.6333 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,216B, BPFP=2.2611 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,188B, BPFP=2.5477 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,904B, BPFP=0.8146 +⌛️ [2/4] FRONTEND: Frontend time: 0.378s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.output: torch.Size([1, 271, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.486s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 271, 128]) + layer.0.v_cache: torch.Size([1, 4, 271, 128]) + layer.1.k_cache: torch.Size([1, 4, 271, 128]) + layer.1.v_cache: torch.Size([1, 4, 271, 128]) + layer.2.k_cache: torch.Size([1, 4, 271, 128]) + layer.2.v_cache: torch.Size([1, 4, 271, 128]) + layer.3.k_cache: torch.Size([1, 4, 271, 128]) + layer.3.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.k_cache: torch.Size([1, 4, 271, 128]) + layer.4.v_cache: torch.Size([1, 4, 271, 128]) + layer.4.output: torch.Size([1, 271, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12629252 22.75831520 + layer.0.v_cache 0.00001451 0.00501919 + layer.1.k_cache 0.49097243 3.11425196 + layer.1.v_cache 0.00000577 0.00222354 + layer.2.k_cache 0.01060468 0.43364268 + layer.2.v_cache 0.00001909 0.00568291 + layer.3.k_cache 0.02748298 1.85947815 + layer.3.v_cache 0.00001876 0.00639479 + layer.4.k_cache 0.00069967 0.11629958 + layer.4.v_cache 0.00005414 0.01248372 + layer.4.output 0.01047103 197.89506458 + ------------------------------------------------------------------------------------- + TOTAL 0.04290952 83.15172022 + (elements=2,358,784) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2358784 +Total Bytes 461728 +BPFP 1.5660 bits/point +EBPFP 3.1320 equivalent bits/point +MSE 83.151720 +---------------------- -------------------------------------------------------- +Time: 0.873s Load: 0.009s, Pack+Encode: 0.378s, Decode+Unpack: 0.486s +---------------------- -------------------------------------------------------- +💾 Converting with 83.1517 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-266.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-266.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-267.zst (52/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-267.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 279, 128) +Output shape: (1, 279, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.output: torch.Size([1, 279, 3584]) -> torch.Size([1, 1, 279, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,596B, BPFP=0.8174 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,252B, BPFP=2.6463 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,576B, BPFP=1.4323 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,992B, BPFP=2.5757 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,772B, BPFP=1.6673 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,260B, BPFP=2.5347 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,920B, BPFP=1.5636 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,960B, BPFP=2.5739 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,812B, BPFP=2.2296 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,580B, BPFP=2.4966 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 95,176B, BPFP=0.7615 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.482s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15022702 22.39655613 + layer.0.v_cache 0.00001395 0.00491695 + layer.1.k_cache 0.54271635 3.06030317 + layer.1.v_cache 0.00000573 0.00208054 + layer.2.k_cache 0.01332666 0.43366608 + layer.2.v_cache 0.00001832 0.00535305 + layer.3.k_cache 0.02579262 1.87110775 + layer.3.v_cache 0.00001838 0.00595422 + layer.4.k_cache 0.00069325 0.11312802 + layer.4.v_cache 0.00005155 0.01194929 + layer.4.output 0.01006869 194.14095302 + ------------------------------------------------------------------------------------- + TOTAL 0.04725557 81.58186390 + (elements=2,428,416) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2428416 +Total Bytes 461896 +BPFP 1.5216 bits/point +EBPFP 3.0433 equivalent bits/point +MSE 81.581864 +---------------------- -------------------------------------------------------- +Time: 0.845s Load: 0.010s, Pack+Encode: 0.352s, Decode+Unpack: 0.482s +---------------------- -------------------------------------------------------- +💾 Converting with 81.5819 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-267.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-267.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-269.zst (53/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-269.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 295, 128) +Output shape: (1, 295, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.output: torch.Size([1, 295, 3584]) -> torch.Size([1, 1, 295, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,312B, BPFP=0.8110 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,700B, BPFP=2.6324 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,988B, BPFP=1.4294 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,508B, BPFP=2.5693 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,340B, BPFP=1.6600 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,792B, BPFP=2.5314 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,588B, BPFP=1.5672 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,532B, BPFP=2.5706 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,752B, BPFP=2.2114 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,960B, BPFP=2.4873 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,712B, BPFP=0.7923 +⌛️ [2/4] FRONTEND: Frontend time: 0.353s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.output: torch.Size([1, 295, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.532s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.output: torch.Size([1, 295, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12416320 22.02248080 + layer.0.v_cache 0.00001370 0.00480798 + layer.1.k_cache 0.62725913 2.97725934 + layer.1.v_cache 0.00000565 0.00207302 + layer.2.k_cache 0.01102555 0.42462701 + layer.2.v_cache 0.00002033 0.00543237 + layer.3.k_cache 0.04118381 1.98847967 + layer.3.v_cache 0.00001949 0.00615935 + layer.4.k_cache 0.00070768 0.11501920 + layer.4.v_cache 0.00005397 0.01219208 + layer.4.output 0.05105524 178.95367736 + ------------------------------------------------------------------------------------- + TOTAL 0.06834348 75.30789837 + (elements=2,567,680) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2567680 +Total Bytes 491184 +BPFP 1.5304 bits/point +EBPFP 3.0607 equivalent bits/point +MSE 75.307898 +---------------------- -------------------------------------------------------- +Time: 0.893s Load: 0.009s, Pack+Encode: 0.353s, Decode+Unpack: 0.532s +---------------------- -------------------------------------------------------- +💾 Converting with 75.3079 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-269.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-269.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-271.zst (54/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-271.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 289, 128) +Output shape: (1, 289, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.0.v_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.1.k_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.1.v_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.2.k_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.2.v_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.3.k_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.3.v_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.4.k_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.4.v_cache: torch.Size([1, 4, 289, 128]) -> torch.Size([1, 1, 289, 512]) + layer.4.output: torch.Size([1, 289, 3584]) -> torch.Size([1, 1, 289, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,648B, BPFP=0.7920 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,488B, BPFP=2.6215 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,020B, BPFP=1.4068 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,324B, BPFP=2.5586 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,408B, BPFP=1.6440 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,576B, BPFP=2.5182 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,800B, BPFP=1.5571 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,132B, BPFP=2.5482 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,452B, BPFP=2.1871 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,536B, BPFP=2.4619 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,752B, BPFP=0.7627 +⌛️ [2/4] FRONTEND: Frontend time: 0.387s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 289, 128]) + layer.0.v_cache: torch.Size([1, 4, 289, 128]) + layer.1.k_cache: torch.Size([1, 4, 289, 128]) + layer.1.v_cache: torch.Size([1, 4, 289, 128]) + layer.2.k_cache: torch.Size([1, 4, 289, 128]) + layer.2.v_cache: torch.Size([1, 4, 289, 128]) + layer.3.k_cache: torch.Size([1, 4, 289, 128]) + layer.3.v_cache: torch.Size([1, 4, 289, 128]) + layer.4.k_cache: torch.Size([1, 4, 289, 128]) + layer.4.v_cache: torch.Size([1, 4, 289, 128]) + layer.4.output: torch.Size([1, 289, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.501s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 289, 128]) + layer.0.v_cache: torch.Size([1, 4, 289, 128]) + layer.1.k_cache: torch.Size([1, 4, 289, 128]) + layer.1.v_cache: torch.Size([1, 4, 289, 128]) + layer.2.k_cache: torch.Size([1, 4, 289, 128]) + layer.2.v_cache: torch.Size([1, 4, 289, 128]) + layer.3.k_cache: torch.Size([1, 4, 289, 128]) + layer.3.v_cache: torch.Size([1, 4, 289, 128]) + layer.4.k_cache: torch.Size([1, 4, 289, 128]) + layer.4.v_cache: torch.Size([1, 4, 289, 128]) + layer.4.output: torch.Size([1, 289, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15516999 22.96456835 + layer.0.v_cache 0.00001361 0.00515022 + layer.1.k_cache 0.56212566 3.01327842 + layer.1.v_cache 0.00000565 0.00227822 + layer.2.k_cache 0.00839357 0.43038352 + layer.2.v_cache 0.00001980 0.00591841 + layer.3.k_cache 0.02402391 1.91140525 + layer.3.v_cache 0.00001987 0.00649867 + layer.4.k_cache 0.00071212 0.11776715 + layer.4.v_cache 0.00005188 0.01260147 + layer.4.output 0.05073151 185.90005561 + ------------------------------------------------------------------------------------- + TOTAL 0.06503863 78.22177876 + (elements=2,515,456) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2515456 +Total Bytes 474136 +BPFP 1.5079 bits/point +EBPFP 3.0158 equivalent bits/point +MSE 78.221779 +---------------------- -------------------------------------------------------- +Time: 0.898s Load: 0.010s, Pack+Encode: 0.387s, Decode+Unpack: 0.501s +---------------------- -------------------------------------------------------- +💾 Converting with 78.2218 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-271.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-271.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-278.zst (55/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-278.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 297, 128) +Output shape: (1, 297, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.output: torch.Size([1, 297, 3584]) -> torch.Size([1, 1, 297, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,672B, BPFP=0.8245 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,004B, BPFP=2.6307 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,252B, BPFP=1.4337 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,792B, BPFP=2.5669 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,668B, BPFP=1.6660 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,200B, BPFP=2.5358 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,852B, BPFP=1.5705 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,004B, BPFP=2.5781 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,048B, BPFP=2.2121 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,192B, BPFP=2.4827 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,904B, BPFP=0.7884 +⌛️ [2/4] FRONTEND: Frontend time: 0.331s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.output: torch.Size([1, 297, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.539s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.output: torch.Size([1, 297, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14005571 22.64716830 + layer.0.v_cache 0.00001459 0.00498926 + layer.1.k_cache 0.64038677 3.03085481 + layer.1.v_cache 0.00000545 0.00204908 + layer.2.k_cache 0.01623841 0.44243213 + layer.2.v_cache 0.00002030 0.00566289 + layer.3.k_cache 0.01774169 1.86722429 + layer.3.v_cache 0.00001948 0.00640722 + layer.4.k_cache 0.00069419 0.11462058 + layer.4.v_cache 0.00005057 0.01207588 + layer.4.output 0.05052170 180.36429774 + ------------------------------------------------------------------------------------- + TOTAL 0.06875759 75.92256286 + (elements=2,585,088) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2585088 +Total Bytes 494588 +BPFP 1.5306 bits/point +EBPFP 3.0612 equivalent bits/point +MSE 75.922563 +---------------------- -------------------------------------------------------- +Time: 0.879s Load: 0.010s, Pack+Encode: 0.331s, Decode+Unpack: 0.539s +---------------------- -------------------------------------------------------- +💾 Converting with 75.9226 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-278.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-278.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-28.zst (56/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-28.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 279, 128) +Output shape: (1, 279, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.output: torch.Size([1, 279, 3584]) -> torch.Size([1, 1, 279, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,600B, BPFP=0.8177 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,348B, BPFP=2.6517 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,572B, BPFP=1.4321 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,264B, BPFP=2.5909 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,008B, BPFP=1.6806 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,556B, BPFP=2.5513 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,100B, BPFP=1.5737 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,188B, BPFP=2.5867 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,692B, BPFP=2.2229 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,744B, BPFP=2.5058 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,180B, BPFP=0.7855 +⌛️ [2/4] FRONTEND: Frontend time: 0.379s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.505s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14010120 22.94441294 + layer.0.v_cache 0.00001437 0.00501874 + layer.1.k_cache 0.58692467 3.10540268 + layer.1.v_cache 0.00000569 0.00208421 + layer.2.k_cache 0.00806839 0.42019172 + layer.2.v_cache 0.00001934 0.00566950 + layer.3.k_cache 0.03832107 2.00138478 + layer.3.v_cache 0.00001992 0.00647580 + layer.4.k_cache 0.00069804 0.11337499 + layer.4.v_cache 0.00005131 0.01233600 + layer.4.output 0.00759059 195.00499232 + ------------------------------------------------------------------------------------- + TOTAL 0.04866812 81.97948809 + (elements=2,428,416) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2428416 +Total Bytes 466252 +BPFP 1.5360 bits/point +EBPFP 3.0720 equivalent bits/point +MSE 81.979488 +---------------------- -------------------------------------------------------- +Time: 0.893s Load: 0.009s, Pack+Encode: 0.379s, Decode+Unpack: 0.505s +---------------------- -------------------------------------------------------- +💾 Converting with 81.9795 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-28.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-28.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-280.zst (57/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-280.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 268, 128) +Output shape: (1, 268, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.output: torch.Size([1, 268, 3584]) -> torch.Size([1, 1, 268, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,252B, BPFP=0.8309 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,620B, BPFP=2.7181 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,012B, BPFP=1.4583 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,484B, BPFP=2.6518 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,172B, BPFP=1.7008 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,744B, BPFP=2.6087 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,136B, BPFP=1.5821 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,364B, BPFP=2.6448 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,220B, BPFP=2.2866 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,972B, BPFP=2.5637 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 95,956B, BPFP=0.7992 +⌛️ [2/4] FRONTEND: Frontend time: 0.359s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.output: torch.Size([1, 268, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.500s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.output: torch.Size([1, 268, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13199043 22.22518511 + layer.0.v_cache 0.00001386 0.00500757 + layer.1.k_cache 0.56989129 3.00567126 + layer.1.v_cache 0.00000577 0.00211697 + layer.2.k_cache 0.00892691 0.43174356 + layer.2.v_cache 0.00001816 0.00557480 + layer.3.k_cache 0.04626510 1.88746552 + layer.3.v_cache 0.00001927 0.00624625 + layer.4.k_cache 0.00069392 0.11798854 + layer.4.v_cache 0.00005195 0.01235177 + layer.4.output 0.00911696 200.96685101 + ------------------------------------------------------------------------------------- + TOTAL 0.04833502 84.38042991 + (elements=2,332,672) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2332672 +Total Bytes 456932 +BPFP 1.5671 bits/point +EBPFP 3.1341 equivalent bits/point +MSE 84.380430 +---------------------- -------------------------------------------------------- +Time: 0.868s Load: 0.009s, Pack+Encode: 0.359s, Decode+Unpack: 0.500s +---------------------- -------------------------------------------------------- +💾 Converting with 84.3804 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-280.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-280.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-285.zst (58/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-285.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 274, 128) +Output shape: (1, 274, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) -> torch.Size([1, 1, 274, 512]) + layer.4.output: torch.Size([1, 274, 3584]) -> torch.Size([1, 1, 274, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,460B, BPFP=0.8246 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,900B, BPFP=2.6745 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,144B, BPFP=1.4339 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,732B, BPFP=2.6079 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,472B, BPFP=1.6807 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,100B, BPFP=2.5719 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,460B, BPFP=1.5659 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,808B, BPFP=2.6122 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,352B, BPFP=2.2441 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,184B, BPFP=2.5196 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 96,572B, BPFP=0.7867 +⌛️ [2/4] FRONTEND: Frontend time: 0.320s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.output: torch.Size([1, 274, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.486s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 274, 128]) + layer.0.v_cache: torch.Size([1, 4, 274, 128]) + layer.1.k_cache: torch.Size([1, 4, 274, 128]) + layer.1.v_cache: torch.Size([1, 4, 274, 128]) + layer.2.k_cache: torch.Size([1, 4, 274, 128]) + layer.2.v_cache: torch.Size([1, 4, 274, 128]) + layer.3.k_cache: torch.Size([1, 4, 274, 128]) + layer.3.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.k_cache: torch.Size([1, 4, 274, 128]) + layer.4.v_cache: torch.Size([1, 4, 274, 128]) + layer.4.output: torch.Size([1, 274, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16500157 22.88907462 + layer.0.v_cache 0.00001391 0.00499909 + layer.1.k_cache 0.57858722 3.09946013 + layer.1.v_cache 0.00000575 0.00210461 + layer.2.k_cache 0.01018238 0.45855771 + layer.2.v_cache 0.00001877 0.00572584 + layer.3.k_cache 0.03425542 2.04588641 + layer.3.v_cache 0.00002021 0.00677471 + layer.4.k_cache 0.00069807 0.11666945 + layer.4.v_cache 0.00005194 0.01245031 + layer.4.output 0.01023613 200.40558199 + ------------------------------------------------------------------------------------- + TOTAL 0.05061695 84.20475158 + (elements=2,384,896) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2384896 +Total Bytes 460184 +BPFP 1.5437 bits/point +EBPFP 3.0873 equivalent bits/point +MSE 84.204752 +---------------------- -------------------------------------------------------- +Time: 0.815s Load: 0.009s, Pack+Encode: 0.320s, Decode+Unpack: 0.486s +---------------------- -------------------------------------------------------- +💾 Converting with 84.2048 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-285.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-285.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-291.zst (59/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-291.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 270, 128) +Output shape: (1, 270, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) -> torch.Size([1, 1, 270, 512]) + layer.4.output: torch.Size([1, 270, 3584]) -> torch.Size([1, 1, 270, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,984B, BPFP=0.8093 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,548B, BPFP=2.6938 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,860B, BPFP=1.4387 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,224B, BPFP=2.6171 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,940B, BPFP=1.6748 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,728B, BPFP=2.5884 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,240B, BPFP=1.5764 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,388B, BPFP=2.6266 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,944B, BPFP=2.2537 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,916B, BPFP=2.5414 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,360B, BPFP=0.8214 +⌛️ [2/4] FRONTEND: Frontend time: 0.338s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.output: torch.Size([1, 270, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.523s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 270, 128]) + layer.0.v_cache: torch.Size([1, 4, 270, 128]) + layer.1.k_cache: torch.Size([1, 4, 270, 128]) + layer.1.v_cache: torch.Size([1, 4, 270, 128]) + layer.2.k_cache: torch.Size([1, 4, 270, 128]) + layer.2.v_cache: torch.Size([1, 4, 270, 128]) + layer.3.k_cache: torch.Size([1, 4, 270, 128]) + layer.3.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.k_cache: torch.Size([1, 4, 270, 128]) + layer.4.v_cache: torch.Size([1, 4, 270, 128]) + layer.4.output: torch.Size([1, 270, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11385483 23.29287652 + layer.0.v_cache 0.00001387 0.00491226 + layer.1.k_cache 0.55871571 3.16481075 + layer.1.v_cache 0.00000565 0.00198244 + layer.2.k_cache 0.00818202 0.43106028 + layer.2.v_cache 0.00001875 0.00576171 + layer.3.k_cache 0.04697755 2.00781521 + layer.3.v_cache 0.00001978 0.00649935 + layer.4.k_cache 0.00071636 0.11319129 + layer.4.v_cache 0.00005146 0.01231688 + layer.4.output 0.01106052 203.93493717 + ------------------------------------------------------------------------------------- + TOTAL 0.04741057 85.68151688 + (elements=2,350,080) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2350080 +Total Bytes 459132 +BPFP 1.5629 bits/point +EBPFP 3.1259 equivalent bits/point +MSE 85.681517 +---------------------- -------------------------------------------------------- +Time: 0.869s Load: 0.008s, Pack+Encode: 0.338s, Decode+Unpack: 0.523s +---------------------- -------------------------------------------------------- +💾 Converting with 85.6815 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-291.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-291.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-292.zst (60/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-292.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 275, 128) +Output shape: (1, 275, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.0.v_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.1.k_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.1.v_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.2.k_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.2.v_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.3.k_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.3.v_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.4.k_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.4.v_cache: torch.Size([1, 4, 275, 128]) -> torch.Size([1, 1, 275, 512]) + layer.4.output: torch.Size([1, 275, 3584]) -> torch.Size([1, 1, 275, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,496B, BPFP=0.8236 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,028B, BPFP=2.6720 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,132B, BPFP=1.4280 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,788B, BPFP=2.6016 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,564B, BPFP=1.6798 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,152B, BPFP=2.5655 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,600B, BPFP=1.5682 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,828B, BPFP=2.6039 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,336B, BPFP=2.2350 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,296B, BPFP=2.5168 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,288B, BPFP=0.7897 +⌛️ [2/4] FRONTEND: Frontend time: 0.336s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 275, 128]) + layer.0.v_cache: torch.Size([1, 4, 275, 128]) + layer.1.k_cache: torch.Size([1, 4, 275, 128]) + layer.1.v_cache: torch.Size([1, 4, 275, 128]) + layer.2.k_cache: torch.Size([1, 4, 275, 128]) + layer.2.v_cache: torch.Size([1, 4, 275, 128]) + layer.3.k_cache: torch.Size([1, 4, 275, 128]) + layer.3.v_cache: torch.Size([1, 4, 275, 128]) + layer.4.k_cache: torch.Size([1, 4, 275, 128]) + layer.4.v_cache: torch.Size([1, 4, 275, 128]) + layer.4.output: torch.Size([1, 275, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.494s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 275, 128]) + layer.0.v_cache: torch.Size([1, 4, 275, 128]) + layer.1.k_cache: torch.Size([1, 4, 275, 128]) + layer.1.v_cache: torch.Size([1, 4, 275, 128]) + layer.2.k_cache: torch.Size([1, 4, 275, 128]) + layer.2.v_cache: torch.Size([1, 4, 275, 128]) + layer.3.k_cache: torch.Size([1, 4, 275, 128]) + layer.3.v_cache: torch.Size([1, 4, 275, 128]) + layer.4.k_cache: torch.Size([1, 4, 275, 128]) + layer.4.v_cache: torch.Size([1, 4, 275, 128]) + layer.4.output: torch.Size([1, 275, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14203608 21.68423118 + layer.0.v_cache 0.00001420 0.00479098 + layer.1.k_cache 0.56673284 3.07817782 + layer.1.v_cache 0.00000552 0.00192472 + layer.2.k_cache 0.00530624 0.42667797 + layer.2.v_cache 0.00001825 0.00547649 + layer.3.k_cache 0.02188354 1.96142423 + layer.3.v_cache 0.00001854 0.00623058 + layer.4.k_cache 0.00069888 0.11257444 + layer.4.v_cache 0.00004918 0.01166592 + layer.4.output 0.00799176 193.75542208 + ------------------------------------------------------------------------------------- + TOTAL 0.04662974 81.38712523 + (elements=2,393,600) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2393600 +Total Bytes 461508 +BPFP 1.5425 bits/point +EBPFP 3.0849 equivalent bits/point +MSE 81.387125 +---------------------- -------------------------------------------------------- +Time: 0.838s Load: 0.009s, Pack+Encode: 0.336s, Decode+Unpack: 0.494s +---------------------- -------------------------------------------------------- +💾 Converting with 81.3871 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-292.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-292.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-30.zst (61/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-30.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,172B, BPFP=0.8377 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,648B, BPFP=2.6307 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,088B, BPFP=1.4404 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,368B, BPFP=2.5601 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,184B, BPFP=1.6665 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,736B, BPFP=2.5252 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,468B, BPFP=1.5718 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,464B, BPFP=2.5654 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,072B, BPFP=2.2125 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,920B, BPFP=2.4801 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 104,492B, BPFP=0.8242 +⌛️ [2/4] FRONTEND: Frontend time: 0.353s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.487s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12732459 22.51241925 + layer.0.v_cache 0.00001480 0.00488516 + layer.1.k_cache 0.57322790 2.96522398 + layer.1.v_cache 0.00000557 0.00210267 + layer.2.k_cache 0.00950604 0.43473549 + layer.2.v_cache 0.00002039 0.00586166 + layer.3.k_cache 0.01510038 1.84107977 + layer.3.v_cache 0.00001916 0.00644584 + layer.4.k_cache 0.00068141 0.11421864 + layer.4.v_cache 0.00005165 0.01214756 + layer.4.output 0.00936729 189.60758455 + ------------------------------------------------------------------------------------- + TOTAL 0.04656017 79.71483599 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 475612 +BPFP 1.5447 bits/point +EBPFP 3.0894 equivalent bits/point +MSE 79.714836 +---------------------- -------------------------------------------------------- +Time: 0.851s Load: 0.010s, Pack+Encode: 0.353s, Decode+Unpack: 0.487s +---------------------- -------------------------------------------------------- +💾 Converting with 79.7148 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-30.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-30.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-300.zst (62/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-300.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 276, 128) +Output shape: (1, 276, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.output: torch.Size([1, 276, 3584]) -> torch.Size([1, 1, 276, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,676B, BPFP=0.8308 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,228B, BPFP=2.6737 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,304B, BPFP=1.4325 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,960B, BPFP=2.6019 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,584B, BPFP=1.6748 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,456B, BPFP=2.5734 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,752B, BPFP=1.5711 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,108B, BPFP=2.6103 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,760B, BPFP=2.2509 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,744B, BPFP=2.5331 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,048B, BPFP=0.8334 +⌛️ [2/4] FRONTEND: Frontend time: 0.337s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.488s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13630490 22.75618490 + layer.0.v_cache 0.00001544 0.00489263 + layer.1.k_cache 0.58776402 3.06624592 + layer.1.v_cache 0.00000559 0.00197072 + layer.2.k_cache 0.01000112 0.43086693 + layer.2.v_cache 0.00001940 0.00546064 + layer.3.k_cache 0.04424953 1.96119004 + layer.3.v_cache 0.00002139 0.00623206 + layer.4.k_cache 0.00071093 0.11469547 + layer.4.v_cache 0.00005158 0.01215129 + layer.4.output 0.01101471 190.24830163 + ------------------------------------------------------------------------------------- + TOTAL 0.05036746 80.00576482 + (elements=2,402,304) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2402304 +Total Bytes 469620 +BPFP 1.5639 bits/point +EBPFP 3.1278 equivalent bits/point +MSE 80.005765 +---------------------- -------------------------------------------------------- +Time: 0.835s Load: 0.009s, Pack+Encode: 0.337s, Decode+Unpack: 0.488s +---------------------- -------------------------------------------------------- +💾 Converting with 80.0058 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-300.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-300.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-308.zst (63/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-308.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 278, 128) +Output shape: (1, 278, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.0.v_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.1.k_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.1.v_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.2.k_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.2.v_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.3.k_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.3.v_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.4.k_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.4.v_cache: torch.Size([1, 4, 278, 128]) -> torch.Size([1, 1, 278, 512]) + layer.4.output: torch.Size([1, 278, 3584]) -> torch.Size([1, 1, 278, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,556B, BPFP=0.8181 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,288B, BPFP=2.6578 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,344B, BPFP=1.4245 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,032B, BPFP=2.5872 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,680B, BPFP=1.6682 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,524B, BPFP=2.5587 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,760B, BPFP=1.5603 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,012B, BPFP=2.5861 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,740B, BPFP=2.2336 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,604B, BPFP=2.5070 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,340B, BPFP=0.7816 +⌛️ [2/4] FRONTEND: Frontend time: 0.334s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 278, 128]) + layer.0.v_cache: torch.Size([1, 4, 278, 128]) + layer.1.k_cache: torch.Size([1, 4, 278, 128]) + layer.1.v_cache: torch.Size([1, 4, 278, 128]) + layer.2.k_cache: torch.Size([1, 4, 278, 128]) + layer.2.v_cache: torch.Size([1, 4, 278, 128]) + layer.3.k_cache: torch.Size([1, 4, 278, 128]) + layer.3.v_cache: torch.Size([1, 4, 278, 128]) + layer.4.k_cache: torch.Size([1, 4, 278, 128]) + layer.4.v_cache: torch.Size([1, 4, 278, 128]) + layer.4.output: torch.Size([1, 278, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.460s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 278, 128]) + layer.0.v_cache: torch.Size([1, 4, 278, 128]) + layer.1.k_cache: torch.Size([1, 4, 278, 128]) + layer.1.v_cache: torch.Size([1, 4, 278, 128]) + layer.2.k_cache: torch.Size([1, 4, 278, 128]) + layer.2.v_cache: torch.Size([1, 4, 278, 128]) + layer.3.k_cache: torch.Size([1, 4, 278, 128]) + layer.3.v_cache: torch.Size([1, 4, 278, 128]) + layer.4.k_cache: torch.Size([1, 4, 278, 128]) + layer.4.v_cache: torch.Size([1, 4, 278, 128]) + layer.4.output: torch.Size([1, 278, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12958164 22.80005234 + layer.0.v_cache 0.00001430 0.00503710 + layer.1.k_cache 0.54121377 3.00856314 + layer.1.v_cache 0.00000577 0.00203143 + layer.2.k_cache 0.01080061 0.42921648 + layer.2.v_cache 0.00001903 0.00570953 + layer.3.k_cache 0.01783211 1.82186297 + layer.3.v_cache 0.00002028 0.00632003 + layer.4.k_cache 0.00068935 0.11466396 + layer.4.v_cache 0.00005053 0.01230681 + layer.4.output 0.00833506 197.32173368 + ------------------------------------------------------------------------------------- + TOTAL 0.04462193 82.90928821 + (elements=2,419,712) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2419712 +Total Bytes 463880 +BPFP 1.5337 bits/point +EBPFP 3.0673 equivalent bits/point +MSE 82.909288 +---------------------- -------------------------------------------------------- +Time: 0.803s Load: 0.009s, Pack+Encode: 0.334s, Decode+Unpack: 0.460s +---------------------- -------------------------------------------------------- +💾 Converting with 82.9093 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-308.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-308.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-309.zst (64/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-309.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 263, 128) +Output shape: (1, 263, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.output: torch.Size([1, 263, 3584]) -> torch.Size([1, 1, 263, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,972B, BPFP=0.8301 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,376B, BPFP=2.7552 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,480B, BPFP=1.4544 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,116B, BPFP=2.6804 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,576B, BPFP=1.6977 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,580B, BPFP=2.6485 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,876B, BPFP=1.5967 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,172B, BPFP=2.6837 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,848B, BPFP=2.3080 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,724B, BPFP=2.5977 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,656B, BPFP=0.8373 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.output: torch.Size([1, 263, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.470s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.output: torch.Size([1, 263, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12547956 23.05815374 + layer.0.v_cache 0.00001492 0.00491439 + layer.1.k_cache 0.50446328 3.16677729 + layer.1.v_cache 0.00000633 0.00206966 + layer.2.k_cache 0.00671430 0.40067715 + layer.2.v_cache 0.00001992 0.00554156 + layer.3.k_cache 0.02142675 1.98599487 + layer.3.v_cache 0.00001986 0.00643341 + layer.4.k_cache 0.00069859 0.11617312 + layer.4.v_cache 0.00005414 0.01249077 + layer.4.output 0.01038299 205.84487032 + ------------------------------------------------------------------------------------- + TOTAL 0.04303403 86.45137166 + (elements=2,289,152) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2289152 +Total Bytes 456376 +BPFP 1.5949 bits/point +EBPFP 3.1898 equivalent bits/point +MSE 86.451372 +---------------------- -------------------------------------------------------- +Time: 0.832s Load: 0.010s, Pack+Encode: 0.352s, Decode+Unpack: 0.470s +---------------------- -------------------------------------------------------- +💾 Converting with 86.4514 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-309.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-309.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-31.zst (65/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-31.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 301, 128) +Output shape: (1, 301, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.output: torch.Size([1, 301, 3584]) -> torch.Size([1, 1, 301, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,396B, BPFP=0.7992 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,304B, BPFP=2.6113 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,548B, BPFP=1.4300 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,020B, BPFP=2.5446 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,812B, BPFP=1.6514 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,408B, BPFP=2.5129 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,980B, BPFP=1.5563 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,016B, BPFP=2.5444 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,100B, BPFP=2.1854 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,356B, BPFP=2.4583 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,684B, BPFP=0.7615 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.output: torch.Size([1, 301, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.514s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.output: torch.Size([1, 301, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13970643 22.45328884 + layer.0.v_cache 0.00001441 0.00492935 + layer.1.k_cache 0.60015088 3.11425741 + layer.1.v_cache 0.00000584 0.00207179 + layer.2.k_cache 0.00648033 0.42796179 + layer.2.v_cache 0.00001840 0.00545403 + layer.3.k_cache 0.03288257 2.03736041 + layer.3.v_cache 0.00002024 0.00631307 + layer.4.k_cache 0.00070287 0.11433125 + layer.4.v_cache 0.00005060 0.01220270 + layer.4.output 0.04874230 178.73240982 + ------------------------------------------------------------------------------------- + TOTAL 0.06595463 75.25323761 + (elements=2,619,904) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2619904 +Total Bytes 493624 +BPFP 1.5073 bits/point +EBPFP 3.0146 equivalent bits/point +MSE 75.253238 +---------------------- -------------------------------------------------------- +Time: 0.875s Load: 0.009s, Pack+Encode: 0.352s, Decode+Unpack: 0.514s +---------------------- -------------------------------------------------------- +💾 Converting with 75.2532 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-31.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-31.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-313.zst (66/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-313.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 327, 128) +Output shape: (1, 327, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.0.v_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.1.k_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.1.v_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.2.k_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.2.v_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.3.k_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.3.v_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.4.k_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.4.v_cache: torch.Size([1, 4, 327, 128]) -> torch.Size([1, 1, 327, 512]) + layer.4.output: torch.Size([1, 327, 3584]) -> torch.Size([1, 1, 327, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 17,356B, BPFP=0.8293 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 56,596B, BPFP=2.7043 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 29,784B, BPFP=1.4232 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 55,076B, BPFP=2.6317 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 34,860B, BPFP=1.6657 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 54,228B, BPFP=2.5912 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 32,884B, BPFP=1.5713 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 55,152B, BPFP=2.6353 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 47,116B, BPFP=2.2513 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 53,176B, BPFP=2.5409 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 114,744B, BPFP=0.7833 +⌛️ [2/4] FRONTEND: Frontend time: 0.388s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 327, 128]) + layer.0.v_cache: torch.Size([1, 4, 327, 128]) + layer.1.k_cache: torch.Size([1, 4, 327, 128]) + layer.1.v_cache: torch.Size([1, 4, 327, 128]) + layer.2.k_cache: torch.Size([1, 4, 327, 128]) + layer.2.v_cache: torch.Size([1, 4, 327, 128]) + layer.3.k_cache: torch.Size([1, 4, 327, 128]) + layer.3.v_cache: torch.Size([1, 4, 327, 128]) + layer.4.k_cache: torch.Size([1, 4, 327, 128]) + layer.4.v_cache: torch.Size([1, 4, 327, 128]) + layer.4.output: torch.Size([1, 327, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.620s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 327, 128]) + layer.0.v_cache: torch.Size([1, 4, 327, 128]) + layer.1.k_cache: torch.Size([1, 4, 327, 128]) + layer.1.v_cache: torch.Size([1, 4, 327, 128]) + layer.2.k_cache: torch.Size([1, 4, 327, 128]) + layer.2.v_cache: torch.Size([1, 4, 327, 128]) + layer.3.k_cache: torch.Size([1, 4, 327, 128]) + layer.3.v_cache: torch.Size([1, 4, 327, 128]) + layer.4.k_cache: torch.Size([1, 4, 327, 128]) + layer.4.v_cache: torch.Size([1, 4, 327, 128]) + layer.4.output: torch.Size([1, 327, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13960611 22.41869147 + layer.0.v_cache 0.00001491 0.00513514 + layer.1.k_cache 0.68890666 3.02301660 + layer.1.v_cache 0.00000583 0.00209478 + layer.2.k_cache 0.00940153 0.44322139 + layer.2.v_cache 0.00001882 0.00553566 + layer.3.k_cache 0.04667800 1.92769051 + layer.3.v_cache 0.00001939 0.00644518 + layer.4.k_cache 0.00070827 0.11192861 + layer.4.v_cache 0.00005291 0.01204802 + layer.4.output 0.04427218 161.49309196 + ------------------------------------------------------------------------------------- + TOTAL 0.07031281 68.14161477 + (elements=2,846,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2846208 +Total Bytes 550972 +BPFP 1.5486 bits/point +EBPFP 3.0973 equivalent bits/point +MSE 68.141615 +---------------------- -------------------------------------------------------- +Time: 1.019s Load: 0.011s, Pack+Encode: 0.388s, Decode+Unpack: 0.620s +---------------------- -------------------------------------------------------- +💾 Converting with 68.1416 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-313.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-313.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-343.zst (67/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-343.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.012s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 299, 128) +Output shape: (1, 299, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) -> torch.Size([1, 1, 299, 512]) + layer.4.output: torch.Size([1, 299, 3584]) -> torch.Size([1, 1, 299, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,172B, BPFP=0.7929 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,124B, BPFP=2.6194 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,436B, BPFP=1.4337 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,856B, BPFP=2.5531 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,748B, BPFP=1.6591 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,212B, BPFP=2.5194 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,824B, BPFP=1.5585 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,776B, BPFP=2.5489 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,120B, BPFP=2.2011 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,256B, BPFP=2.4695 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 105,648B, BPFP=0.7887 +⌛️ [2/4] FRONTEND: Frontend time: 0.371s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.output: torch.Size([1, 299, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.502s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 299, 128]) + layer.0.v_cache: torch.Size([1, 4, 299, 128]) + layer.1.k_cache: torch.Size([1, 4, 299, 128]) + layer.1.v_cache: torch.Size([1, 4, 299, 128]) + layer.2.k_cache: torch.Size([1, 4, 299, 128]) + layer.2.v_cache: torch.Size([1, 4, 299, 128]) + layer.3.k_cache: torch.Size([1, 4, 299, 128]) + layer.3.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.k_cache: torch.Size([1, 4, 299, 128]) + layer.4.v_cache: torch.Size([1, 4, 299, 128]) + layer.4.output: torch.Size([1, 299, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15817952 21.92704686 + layer.0.v_cache 0.00001474 0.00486849 + layer.1.k_cache 0.61609703 2.90747662 + layer.1.v_cache 0.00000566 0.00205214 + layer.2.k_cache 0.00844678 0.44103810 + layer.2.v_cache 0.00001876 0.00548204 + layer.3.k_cache 0.04682518 1.92189133 + layer.3.v_cache 0.00001896 0.00616857 + layer.4.k_cache 0.00069859 0.11282472 + layer.4.v_cache 0.00005216 0.01218273 + layer.4.output 0.04649196 175.78374343 + ------------------------------------------------------------------------------------- + TOTAL 0.06798830 73.98983739 + (elements=2,602,496) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2602496 +Total Bytes 495172 +BPFP 1.5221 bits/point +EBPFP 3.0443 equivalent bits/point +MSE 73.989837 +---------------------- -------------------------------------------------------- +Time: 0.885s Load: 0.012s, Pack+Encode: 0.371s, Decode+Unpack: 0.502s +---------------------- -------------------------------------------------------- +💾 Converting with 73.9898 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-343.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-343.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-344.zst (68/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-344.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 287, 128) +Output shape: (1, 287, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.output: torch.Size([1, 287, 3584]) -> torch.Size([1, 1, 287, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,916B, BPFP=0.8121 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,504B, BPFP=2.6407 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,384B, BPFP=1.4364 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,308B, BPFP=2.5756 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,700B, BPFP=1.6714 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,656B, BPFP=2.5401 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,984B, BPFP=1.5780 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,180B, BPFP=2.5686 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,772B, BPFP=2.2197 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,504B, BPFP=2.4774 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,892B, BPFP=0.8080 +⌛️ [2/4] FRONTEND: Frontend time: 0.364s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.output: torch.Size([1, 287, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.477s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.output: torch.Size([1, 287, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14002854 23.19129736 + layer.0.v_cache 0.00001601 0.00501226 + layer.1.k_cache 0.54217949 2.97769388 + layer.1.v_cache 0.00000579 0.00209671 + layer.2.k_cache 0.01082110 0.43906586 + layer.2.v_cache 0.00001971 0.00567592 + layer.3.k_cache 0.03102738 1.83122758 + layer.3.v_cache 0.00001907 0.00629793 + layer.4.k_cache 0.00066782 0.11524854 + layer.4.v_cache 0.00005141 0.01225845 + layer.4.output 0.00856273 187.41494525 + ------------------------------------------------------------------------------------- + TOTAL 0.04616326 78.85238184 + (elements=2,498,048) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2498048 +Total Bytes 480800 +BPFP 1.5398 bits/point +EBPFP 3.0795 equivalent bits/point +MSE 78.852382 +---------------------- -------------------------------------------------------- +Time: 0.851s Load: 0.010s, Pack+Encode: 0.364s, Decode+Unpack: 0.477s +---------------------- -------------------------------------------------------- +💾 Converting with 78.8524 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-344.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-344.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-350.zst (69/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-350.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 297, 128) +Output shape: (1, 297, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) -> torch.Size([1, 1, 297, 512]) + layer.4.output: torch.Size([1, 297, 3584]) -> torch.Size([1, 1, 297, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,804B, BPFP=0.8314 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,860B, BPFP=2.6231 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,116B, BPFP=1.4266 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,632B, BPFP=2.5585 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,556B, BPFP=1.6601 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,056B, BPFP=2.5282 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,852B, BPFP=1.5705 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,792B, BPFP=2.5669 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,812B, BPFP=2.1997 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,028B, BPFP=2.4741 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,212B, BPFP=0.7757 +⌛️ [2/4] FRONTEND: Frontend time: 0.324s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.output: torch.Size([1, 297, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.561s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 297, 128]) + layer.0.v_cache: torch.Size([1, 4, 297, 128]) + layer.1.k_cache: torch.Size([1, 4, 297, 128]) + layer.1.v_cache: torch.Size([1, 4, 297, 128]) + layer.2.k_cache: torch.Size([1, 4, 297, 128]) + layer.2.v_cache: torch.Size([1, 4, 297, 128]) + layer.3.k_cache: torch.Size([1, 4, 297, 128]) + layer.3.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.k_cache: torch.Size([1, 4, 297, 128]) + layer.4.v_cache: torch.Size([1, 4, 297, 128]) + layer.4.output: torch.Size([1, 297, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14705232 22.51777541 + layer.0.v_cache 0.00001374 0.00486056 + layer.1.k_cache 0.65666707 3.03130014 + layer.1.v_cache 0.00000579 0.00204212 + layer.2.k_cache 0.01062412 0.45433795 + layer.2.v_cache 0.00001918 0.00550030 + layer.3.k_cache 0.03205616 2.01740879 + layer.3.v_cache 0.00001999 0.00648525 + layer.4.k_cache 0.00071739 0.11474843 + layer.4.v_cache 0.00005111 0.01173937 + layer.4.output 0.04976816 179.99467893 + ------------------------------------------------------------------------------------- + TOTAL 0.07032965 75.77229123 + (elements=2,585,088) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2585088 +Total Bytes 491720 +BPFP 1.5217 bits/point +EBPFP 3.0434 equivalent bits/point +MSE 75.772291 +---------------------- -------------------------------------------------------- +Time: 0.895s Load: 0.010s, Pack+Encode: 0.324s, Decode+Unpack: 0.561s +---------------------- -------------------------------------------------------- +💾 Converting with 75.7723 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-350.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-350.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-351.zst (70/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-351.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 281, 128) +Output shape: (1, 281, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) -> torch.Size([1, 1, 281, 512]) + layer.4.output: torch.Size([1, 281, 3584]) -> torch.Size([1, 1, 281, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,096B, BPFP=0.8394 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,464B, BPFP=2.6392 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,656B, BPFP=1.4266 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,176B, BPFP=2.5676 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,976B, BPFP=1.6668 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,688B, BPFP=2.5405 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,088B, BPFP=1.5618 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,252B, BPFP=2.5718 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,800B, BPFP=2.2131 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,672B, BPFP=2.4840 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,884B, BPFP=0.8014 +⌛️ [2/4] FRONTEND: Frontend time: 0.342s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.output: torch.Size([1, 281, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.484s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 281, 128]) + layer.0.v_cache: torch.Size([1, 4, 281, 128]) + layer.1.k_cache: torch.Size([1, 4, 281, 128]) + layer.1.v_cache: torch.Size([1, 4, 281, 128]) + layer.2.k_cache: torch.Size([1, 4, 281, 128]) + layer.2.v_cache: torch.Size([1, 4, 281, 128]) + layer.3.k_cache: torch.Size([1, 4, 281, 128]) + layer.3.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.k_cache: torch.Size([1, 4, 281, 128]) + layer.4.v_cache: torch.Size([1, 4, 281, 128]) + layer.4.output: torch.Size([1, 281, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15131647 23.18167190 + layer.0.v_cache 0.00001392 0.00484860 + layer.1.k_cache 0.51967005 3.07564580 + layer.1.v_cache 0.00000557 0.00193952 + layer.2.k_cache 0.01013049 0.43872526 + layer.2.v_cache 0.00001908 0.00532778 + layer.3.k_cache 0.05186962 2.06538929 + layer.3.v_cache 0.00001891 0.00608744 + layer.4.k_cache 0.00069958 0.11470032 + layer.4.v_cache 0.00005017 0.01171072 + layer.4.output 0.01073789 195.42253432 + ------------------------------------------------------------------------------------- + TOTAL 0.04758583 82.16845805 + (elements=2,445,824) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2445824 +Total Bytes 469752 +BPFP 1.5365 bits/point +EBPFP 3.0730 equivalent bits/point +MSE 82.168458 +---------------------- -------------------------------------------------------- +Time: 0.835s Load: 0.009s, Pack+Encode: 0.342s, Decode+Unpack: 0.484s +---------------------- -------------------------------------------------------- +💾 Converting with 82.1685 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-351.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-351.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-360.zst (71/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-360.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 288, 128) +Output shape: (1, 288, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) -> torch.Size([1, 1, 288, 512]) + layer.4.output: torch.Size([1, 288, 3584]) -> torch.Size([1, 1, 288, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,644B, BPFP=0.7945 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,332B, BPFP=2.6222 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,084B, BPFP=1.4151 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,148B, BPFP=2.5579 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,480B, BPFP=1.6536 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,540B, BPFP=2.5250 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,036B, BPFP=1.5753 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,148B, BPFP=2.5579 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,680B, BPFP=2.2070 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,480B, BPFP=2.4674 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,100B, BPFP=0.7836 +⌛️ [2/4] FRONTEND: Frontend time: 0.384s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.510s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 288, 128]) + layer.0.v_cache: torch.Size([1, 4, 288, 128]) + layer.1.k_cache: torch.Size([1, 4, 288, 128]) + layer.1.v_cache: torch.Size([1, 4, 288, 128]) + layer.2.k_cache: torch.Size([1, 4, 288, 128]) + layer.2.v_cache: torch.Size([1, 4, 288, 128]) + layer.3.k_cache: torch.Size([1, 4, 288, 128]) + layer.3.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.k_cache: torch.Size([1, 4, 288, 128]) + layer.4.v_cache: torch.Size([1, 4, 288, 128]) + layer.4.output: torch.Size([1, 288, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15566496 22.20855713 + layer.0.v_cache 0.00001437 0.00478763 + layer.1.k_cache 0.62643353 3.01129341 + layer.1.v_cache 0.00000573 0.00203097 + layer.2.k_cache 0.01010078 0.43748040 + layer.2.v_cache 0.00001868 0.00550875 + layer.3.k_cache 0.06271097 1.85448922 + layer.3.v_cache 0.00001958 0.00629090 + layer.4.k_cache 0.00073784 0.11333801 + layer.4.v_cache 0.00005294 0.01202236 + layer.4.output 0.00677609 188.67004588 + ------------------------------------------------------------------------------------- + TOTAL 0.05312894 79.31447764 + (elements=2,506,752) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2506752 +Total Bytes 476672 +BPFP 1.5212 bits/point +EBPFP 3.0425 equivalent bits/point +MSE 79.314478 +---------------------- -------------------------------------------------------- +Time: 0.904s Load: 0.009s, Pack+Encode: 0.384s, Decode+Unpack: 0.510s +---------------------- -------------------------------------------------------- +💾 Converting with 79.3145 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-360.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-360.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-367.zst (72/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-367.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 286, 128) +Output shape: (1, 286, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.output: torch.Size([1, 286, 3584]) -> torch.Size([1, 1, 286, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,088B, BPFP=0.8243 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,052B, BPFP=2.6252 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,436B, BPFP=1.4443 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,936B, BPFP=2.5642 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,692B, BPFP=1.6768 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,420B, BPFP=2.5361 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,912B, BPFP=1.5795 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,816B, BPFP=2.5577 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,624B, BPFP=2.2194 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,452B, BPFP=2.4832 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,604B, BPFP=0.8008 +⌛️ [2/4] FRONTEND: Frontend time: 0.366s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.491s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13918158 23.33582141 + layer.0.v_cache 0.00001566 0.00502669 + layer.1.k_cache 0.57997750 2.92313860 + layer.1.v_cache 0.00000571 0.00208000 + layer.2.k_cache 0.01637170 0.45554891 + layer.2.v_cache 0.00001887 0.00589644 + layer.3.k_cache 0.04393929 1.90479433 + layer.3.v_cache 0.00001903 0.00631343 + layer.4.k_cache 0.00071104 0.11606174 + layer.4.v_cache 0.00005175 0.01267251 + layer.4.output 0.01065216 192.93373814 + ------------------------------------------------------------------------------------- + TOTAL 0.05028572 81.13550124 + (elements=2,489,344) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2489344 +Total Bytes 478032 +BPFP 1.5363 bits/point +EBPFP 3.0725 equivalent bits/point +MSE 81.135501 +---------------------- -------------------------------------------------------- +Time: 0.866s Load: 0.009s, Pack+Encode: 0.366s, Decode+Unpack: 0.491s +---------------------- -------------------------------------------------------- +💾 Converting with 81.1355 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-367.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-367.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-375.zst (73/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-375.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 310, 128) +Output shape: (1, 310, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.0.v_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.1.k_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.1.v_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.2.k_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.2.v_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.3.k_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.3.v_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.4.k_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.4.v_cache: torch.Size([1, 4, 310, 128]) -> torch.Size([1, 1, 310, 512]) + layer.4.output: torch.Size([1, 310, 3584]) -> torch.Size([1, 1, 310, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,792B, BPFP=0.7960 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,972B, BPFP=2.5692 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,768B, BPFP=1.3996 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,748B, BPFP=2.5075 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,416B, BPFP=1.6339 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 49,040B, BPFP=2.4718 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,660B, BPFP=1.5454 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,720B, BPFP=2.5060 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,988B, BPFP=2.1667 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 48,120B, BPFP=2.4254 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 105,560B, BPFP=0.7601 +⌛️ [2/4] FRONTEND: Frontend time: 0.349s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 310, 128]) + layer.0.v_cache: torch.Size([1, 4, 310, 128]) + layer.1.k_cache: torch.Size([1, 4, 310, 128]) + layer.1.v_cache: torch.Size([1, 4, 310, 128]) + layer.2.k_cache: torch.Size([1, 4, 310, 128]) + layer.2.v_cache: torch.Size([1, 4, 310, 128]) + layer.3.k_cache: torch.Size([1, 4, 310, 128]) + layer.3.v_cache: torch.Size([1, 4, 310, 128]) + layer.4.k_cache: torch.Size([1, 4, 310, 128]) + layer.4.v_cache: torch.Size([1, 4, 310, 128]) + layer.4.output: torch.Size([1, 310, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.537s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 310, 128]) + layer.0.v_cache: torch.Size([1, 4, 310, 128]) + layer.1.k_cache: torch.Size([1, 4, 310, 128]) + layer.1.v_cache: torch.Size([1, 4, 310, 128]) + layer.2.k_cache: torch.Size([1, 4, 310, 128]) + layer.2.v_cache: torch.Size([1, 4, 310, 128]) + layer.3.k_cache: torch.Size([1, 4, 310, 128]) + layer.3.v_cache: torch.Size([1, 4, 310, 128]) + layer.4.k_cache: torch.Size([1, 4, 310, 128]) + layer.4.v_cache: torch.Size([1, 4, 310, 128]) + layer.4.output: torch.Size([1, 310, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16496690 22.49384451 + layer.0.v_cache 0.00001433 0.00501802 + layer.1.k_cache 0.65265572 3.18961615 + layer.1.v_cache 0.00000578 0.00207525 + layer.2.k_cache 0.01241827 0.43798365 + layer.2.v_cache 0.00001920 0.00562472 + layer.3.k_cache 0.08444468 1.91137873 + layer.3.v_cache 0.00001975 0.00650078 + layer.4.k_cache 0.00068928 0.11447830 + layer.4.v_cache 0.00005259 0.01234915 + layer.4.output 0.04785494 172.04324597 + ------------------------------------------------------------------------------------- + TOTAL 0.07354536 72.49891712 + (elements=2,698,240) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2698240 +Total Bytes 502784 +BPFP 1.4907 bits/point +EBPFP 2.9814 equivalent bits/point +MSE 72.498917 +---------------------- -------------------------------------------------------- +Time: 0.897s Load: 0.011s, Pack+Encode: 0.349s, Decode+Unpack: 0.537s +---------------------- -------------------------------------------------------- +💾 Converting with 72.4989 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-375.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-375.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-38.zst (74/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-38.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,292B, BPFP=0.8127 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,612B, BPFP=2.6367 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,956B, BPFP=1.4326 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,372B, BPFP=2.5708 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,288B, BPFP=1.6628 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,692B, BPFP=2.5347 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,416B, BPFP=1.5634 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,188B, BPFP=2.5610 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,688B, BPFP=2.2156 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,744B, BPFP=2.4843 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,448B, BPFP=0.7778 +⌛️ [2/4] FRONTEND: Frontend time: 0.400s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.532s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15001445 22.93546383 + layer.0.v_cache 0.00001441 0.00498798 + layer.1.k_cache 0.57475359 2.97346694 + layer.1.v_cache 0.00000584 0.00218294 + layer.2.k_cache 0.01434258 0.46144031 + layer.2.v_cache 0.00001954 0.00582012 + layer.3.k_cache 0.02649183 1.88639697 + layer.3.v_cache 0.00001945 0.00635422 + layer.4.k_cache 0.00069414 0.11911170 + layer.4.v_cache 0.00005306 0.01277611 + layer.4.output 0.05035525 182.45714893 + ------------------------------------------------------------------------------------- + TOTAL 0.06581739 76.80047316 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 487696 +BPFP 1.5247 bits/point +EBPFP 3.0493 equivalent bits/point +MSE 76.800473 +---------------------- -------------------------------------------------------- +Time: 0.941s Load: 0.010s, Pack+Encode: 0.400s, Decode+Unpack: 0.532s +---------------------- -------------------------------------------------------- +💾 Converting with 76.8005 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-38.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-38.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-386.zst (75/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-386.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 282, 128) +Output shape: (1, 282, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) -> torch.Size([1, 1, 282, 512]) + layer.4.output: torch.Size([1, 282, 3584]) -> torch.Size([1, 1, 282, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,844B, BPFP=0.8225 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,696B, BPFP=2.6427 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,892B, BPFP=1.4346 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,416B, BPFP=2.5718 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,352B, BPFP=1.6817 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,824B, BPFP=2.5390 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,376B, BPFP=1.5723 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,416B, BPFP=2.5718 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,336B, BPFP=2.2349 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,020B, BPFP=2.4945 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,664B, BPFP=0.7968 +⌛️ [2/4] FRONTEND: Frontend time: 0.352s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.output: torch.Size([1, 282, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.468s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 282, 128]) + layer.0.v_cache: torch.Size([1, 4, 282, 128]) + layer.1.k_cache: torch.Size([1, 4, 282, 128]) + layer.1.v_cache: torch.Size([1, 4, 282, 128]) + layer.2.k_cache: torch.Size([1, 4, 282, 128]) + layer.2.v_cache: torch.Size([1, 4, 282, 128]) + layer.3.k_cache: torch.Size([1, 4, 282, 128]) + layer.3.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.k_cache: torch.Size([1, 4, 282, 128]) + layer.4.v_cache: torch.Size([1, 4, 282, 128]) + layer.4.output: torch.Size([1, 282, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11854591 22.90817542 + layer.0.v_cache 0.00001566 0.00489479 + layer.1.k_cache 0.56985614 2.98411441 + layer.1.v_cache 0.00000555 0.00207541 + layer.2.k_cache 0.01267573 0.46296578 + layer.2.v_cache 0.00001902 0.00562756 + layer.3.k_cache 0.04477839 1.97341529 + layer.3.v_cache 0.00001896 0.00644016 + layer.4.k_cache 0.00071754 0.11599389 + layer.4.v_cache 0.00005321 0.01233780 + layer.4.output 0.00761333 194.60768110 + ------------------------------------------------------------------------------------- + TOTAL 0.04705761 81.80763578 + (elements=2,454,528) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2454528 +Total Bytes 471836 +BPFP 1.5378 bits/point +EBPFP 3.0757 equivalent bits/point +MSE 81.807636 +---------------------- -------------------------------------------------------- +Time: 0.828s Load: 0.009s, Pack+Encode: 0.352s, Decode+Unpack: 0.468s +---------------------- -------------------------------------------------------- +💾 Converting with 81.8076 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-386.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-386.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-402.zst (76/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-402.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,180B, BPFP=0.8068 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,740B, BPFP=2.6435 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,944B, BPFP=1.4320 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,444B, BPFP=2.5746 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,444B, BPFP=1.6711 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,844B, BPFP=2.5427 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,752B, BPFP=1.5812 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,480B, BPFP=2.5765 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,956B, BPFP=2.2298 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,912B, BPFP=2.4932 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,144B, BPFP=0.7755 +⌛️ [2/4] FRONTEND: Frontend time: 0.357s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.455s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11582189 23.03537880 + layer.0.v_cache 0.00001509 0.00515450 + layer.1.k_cache 0.58311327 3.02089852 + layer.1.v_cache 0.00000606 0.00217121 + layer.2.k_cache 0.00657600 0.41332909 + layer.2.v_cache 0.00001909 0.00580958 + layer.3.k_cache 0.02234380 2.09251310 + layer.3.v_cache 0.00001905 0.00656473 + layer.4.k_cache 0.00071828 0.11732351 + layer.4.v_cache 0.00005370 0.01269623 + layer.4.output 0.05044473 182.87017128 + ------------------------------------------------------------------------------------- + TOTAL 0.06363526 76.98841402 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 488840 +BPFP 1.5282 bits/point +EBPFP 3.0565 equivalent bits/point +MSE 76.988414 +---------------------- -------------------------------------------------------- +Time: 0.821s Load: 0.009s, Pack+Encode: 0.357s, Decode+Unpack: 0.455s +---------------------- -------------------------------------------------------- +💾 Converting with 76.9884 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-402.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-402.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-413.zst (77/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-413.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 292, 128) +Output shape: (1, 292, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.0.v_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.1.k_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.1.v_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.2.k_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.2.v_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.3.k_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.3.v_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.4.k_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.4.v_cache: torch.Size([1, 4, 292, 128]) -> torch.Size([1, 1, 292, 512]) + layer.4.output: torch.Size([1, 292, 3584]) -> torch.Size([1, 1, 292, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,272B, BPFP=0.8172 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,396B, BPFP=2.6432 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,604B, BPFP=1.4236 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,224B, BPFP=2.5805 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,436B, BPFP=1.6821 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,556B, BPFP=2.5447 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,528B, BPFP=1.5801 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,308B, BPFP=2.5850 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,608B, BPFP=2.2265 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,564B, BPFP=2.4917 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,600B, BPFP=0.7614 +⌛️ [2/4] FRONTEND: Frontend time: 0.337s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 292, 128]) + layer.0.v_cache: torch.Size([1, 4, 292, 128]) + layer.1.k_cache: torch.Size([1, 4, 292, 128]) + layer.1.v_cache: torch.Size([1, 4, 292, 128]) + layer.2.k_cache: torch.Size([1, 4, 292, 128]) + layer.2.v_cache: torch.Size([1, 4, 292, 128]) + layer.3.k_cache: torch.Size([1, 4, 292, 128]) + layer.3.v_cache: torch.Size([1, 4, 292, 128]) + layer.4.k_cache: torch.Size([1, 4, 292, 128]) + layer.4.v_cache: torch.Size([1, 4, 292, 128]) + layer.4.output: torch.Size([1, 292, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.492s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 292, 128]) + layer.0.v_cache: torch.Size([1, 4, 292, 128]) + layer.1.k_cache: torch.Size([1, 4, 292, 128]) + layer.1.v_cache: torch.Size([1, 4, 292, 128]) + layer.2.k_cache: torch.Size([1, 4, 292, 128]) + layer.2.v_cache: torch.Size([1, 4, 292, 128]) + layer.3.k_cache: torch.Size([1, 4, 292, 128]) + layer.3.v_cache: torch.Size([1, 4, 292, 128]) + layer.4.k_cache: torch.Size([1, 4, 292, 128]) + layer.4.v_cache: torch.Size([1, 4, 292, 128]) + layer.4.output: torch.Size([1, 292, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14887579 22.46344579 + layer.0.v_cache 0.00001384 0.00489481 + layer.1.k_cache 0.57277983 3.05775305 + layer.1.v_cache 0.00000568 0.00205960 + layer.2.k_cache 0.01314942 0.44524587 + layer.2.v_cache 0.00002105 0.00551366 + layer.3.k_cache 0.03125376 1.99641899 + layer.3.v_cache 0.00001978 0.00626502 + layer.4.k_cache 0.00068035 0.11060026 + layer.4.v_cache 0.00005123 0.01197656 + layer.4.output 0.04859251 179.14330051 + ------------------------------------------------------------------------------------- + TOTAL 0.06511755 75.41807513 + (elements=2,541,568) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2541568 +Total Bytes 484096 +BPFP 1.5238 bits/point +EBPFP 3.0475 equivalent bits/point +MSE 75.418075 +---------------------- -------------------------------------------------------- +Time: 0.838s Load: 0.009s, Pack+Encode: 0.337s, Decode+Unpack: 0.492s +---------------------- -------------------------------------------------------- +💾 Converting with 75.4181 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-413.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-413.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-417.zst (78/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-417.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 338, 128) +Output shape: (1, 338, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.0.v_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.1.k_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.1.v_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.2.k_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.2.v_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.3.k_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.3.v_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.4.k_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.4.v_cache: torch.Size([1, 4, 338, 128]) -> torch.Size([1, 1, 338, 512]) + layer.4.output: torch.Size([1, 338, 3584]) -> torch.Size([1, 1, 338, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 17,756B, BPFP=0.8208 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 57,336B, BPFP=2.6505 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 30,400B, BPFP=1.4053 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 55,856B, BPFP=2.5821 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 35,756B, BPFP=1.6529 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 54,944B, BPFP=2.5399 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 33,504B, BPFP=1.5488 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 55,680B, BPFP=2.5740 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 47,744B, BPFP=2.2071 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 54,004B, BPFP=2.4965 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 122,784B, BPFP=0.8109 +⌛️ [2/4] FRONTEND: Frontend time: 0.416s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 338, 128]) + layer.0.v_cache: torch.Size([1, 4, 338, 128]) + layer.1.k_cache: torch.Size([1, 4, 338, 128]) + layer.1.v_cache: torch.Size([1, 4, 338, 128]) + layer.2.k_cache: torch.Size([1, 4, 338, 128]) + layer.2.v_cache: torch.Size([1, 4, 338, 128]) + layer.3.k_cache: torch.Size([1, 4, 338, 128]) + layer.3.v_cache: torch.Size([1, 4, 338, 128]) + layer.4.k_cache: torch.Size([1, 4, 338, 128]) + layer.4.v_cache: torch.Size([1, 4, 338, 128]) + layer.4.output: torch.Size([1, 338, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.569s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 338, 128]) + layer.0.v_cache: torch.Size([1, 4, 338, 128]) + layer.1.k_cache: torch.Size([1, 4, 338, 128]) + layer.1.v_cache: torch.Size([1, 4, 338, 128]) + layer.2.k_cache: torch.Size([1, 4, 338, 128]) + layer.2.v_cache: torch.Size([1, 4, 338, 128]) + layer.3.k_cache: torch.Size([1, 4, 338, 128]) + layer.3.v_cache: torch.Size([1, 4, 338, 128]) + layer.4.k_cache: torch.Size([1, 4, 338, 128]) + layer.4.v_cache: torch.Size([1, 4, 338, 128]) + layer.4.output: torch.Size([1, 338, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13976314 22.24140163 + layer.0.v_cache 0.00001472 0.00501985 + layer.1.k_cache 0.73750278 2.96395296 + layer.1.v_cache 0.00000589 0.00208669 + layer.2.k_cache 0.02011468 0.43712797 + layer.2.v_cache 0.00002041 0.00539894 + layer.3.k_cache 0.02722352 1.82439969 + layer.3.v_cache 0.00002029 0.00614571 + layer.4.k_cache 0.00072609 0.11691579 + layer.4.v_cache 0.00005477 0.01251323 + layer.4.output 0.04270584 157.47140480 + ------------------------------------------------------------------------------------- + TOTAL 0.07202277 66.46557624 + (elements=2,941,952) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2941952 +Total Bytes 565764 +BPFP 1.5385 bits/point +EBPFP 3.0769 equivalent bits/point +MSE 66.465576 +---------------------- -------------------------------------------------------- +Time: 0.995s Load: 0.010s, Pack+Encode: 0.416s, Decode+Unpack: 0.569s +---------------------- -------------------------------------------------------- +💾 Converting with 66.4656 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-417.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-417.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-42.zst (79/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-42.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 291, 128) +Output shape: (1, 291, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.output: torch.Size([1, 291, 3584]) -> torch.Size([1, 1, 291, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,096B, BPFP=0.8106 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,200B, BPFP=2.6418 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,628B, BPFP=1.4298 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,968B, BPFP=2.5756 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,972B, BPFP=1.6630 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,124B, BPFP=2.5303 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,272B, BPFP=1.5717 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,668B, BPFP=2.5595 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,432B, BPFP=2.2247 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,192B, BPFP=2.4802 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,892B, BPFP=0.7509 +⌛️ [2/4] FRONTEND: Frontend time: 0.351s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.output: torch.Size([1, 291, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.469s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.output: torch.Size([1, 291, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13150811 22.52568091 + layer.0.v_cache 0.00001374 0.00491245 + layer.1.k_cache 0.66021251 3.09931846 + layer.1.v_cache 0.00000574 0.00205054 + layer.2.k_cache 0.00963001 0.43067924 + layer.2.v_cache 0.00001988 0.00564131 + layer.3.k_cache 0.08055985 1.97194312 + layer.3.v_cache 0.00001942 0.00635871 + layer.4.k_cache 0.00069026 0.11505742 + layer.4.v_cache 0.00005126 0.01218713 + layer.4.output 0.05093874 178.95775037 + ------------------------------------------------------------------------------------- + TOTAL 0.07289894 75.34576952 + (elements=2,532,864) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2532864 +Total Bytes 479444 +BPFP 1.5143 bits/point +EBPFP 3.0286 equivalent bits/point +MSE 75.345770 +---------------------- -------------------------------------------------------- +Time: 0.830s Load: 0.009s, Pack+Encode: 0.351s, Decode+Unpack: 0.469s +---------------------- -------------------------------------------------------- +💾 Converting with 75.3458 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-42.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-42.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-429.zst (80/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-429.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 303, 128) +Output shape: (1, 303, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) -> torch.Size([1, 1, 303, 512]) + layer.4.output: torch.Size([1, 303, 3584]) -> torch.Size([1, 1, 303, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,628B, BPFP=0.8059 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,280B, BPFP=2.5928 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,508B, BPFP=1.4185 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,032B, BPFP=2.5285 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,840B, BPFP=1.6419 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,484B, BPFP=2.5002 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,984B, BPFP=1.5462 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,044B, BPFP=2.5291 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,272B, BPFP=2.1799 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,432B, BPFP=2.4460 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 107,600B, BPFP=0.7927 +⌛️ [2/4] FRONTEND: Frontend time: 0.346s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.output: torch.Size([1, 303, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.483s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 303, 128]) + layer.0.v_cache: torch.Size([1, 4, 303, 128]) + layer.1.k_cache: torch.Size([1, 4, 303, 128]) + layer.1.v_cache: torch.Size([1, 4, 303, 128]) + layer.2.k_cache: torch.Size([1, 4, 303, 128]) + layer.2.v_cache: torch.Size([1, 4, 303, 128]) + layer.3.k_cache: torch.Size([1, 4, 303, 128]) + layer.3.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.k_cache: torch.Size([1, 4, 303, 128]) + layer.4.v_cache: torch.Size([1, 4, 303, 128]) + layer.4.output: torch.Size([1, 303, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14893842 22.20986747 + layer.0.v_cache 0.00001415 0.00471432 + layer.1.k_cache 0.65110169 3.06581745 + layer.1.v_cache 0.00000575 0.00208364 + layer.2.k_cache 0.01411552 0.44141944 + layer.2.v_cache 0.00001904 0.00540246 + layer.3.k_cache 0.03939451 1.85442410 + layer.3.v_cache 0.00001876 0.00593459 + layer.4.k_cache 0.00070558 0.11305733 + layer.4.v_cache 0.00005345 0.01191298 + layer.4.output 0.04819407 172.54670556 + ------------------------------------------------------------------------------------- + TOTAL 0.07010149 72.67891604 + (elements=2,637,312) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2637312 +Total Bytes 499104 +BPFP 1.5140 bits/point +EBPFP 3.0280 equivalent bits/point +MSE 72.678916 +---------------------- -------------------------------------------------------- +Time: 0.838s Load: 0.010s, Pack+Encode: 0.346s, Decode+Unpack: 0.483s +---------------------- -------------------------------------------------------- +💾 Converting with 72.6789 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-429.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-429.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-431.zst (81/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-431.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 263, 128) +Output shape: (1, 263, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) -> torch.Size([1, 1, 263, 512]) + layer.4.output: torch.Size([1, 263, 3584]) -> torch.Size([1, 1, 263, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 13,772B, BPFP=0.8182 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,216B, BPFP=2.7457 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,332B, BPFP=1.4456 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,032B, BPFP=2.6754 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,528B, BPFP=1.6949 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,376B, BPFP=2.6364 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,848B, BPFP=1.5951 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,140B, BPFP=2.6818 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,580B, BPFP=2.2921 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,440B, BPFP=2.5808 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 93,056B, BPFP=0.7898 +⌛️ [2/4] FRONTEND: Frontend time: 0.362s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.output: torch.Size([1, 263, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.490s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 263, 128]) + layer.0.v_cache: torch.Size([1, 4, 263, 128]) + layer.1.k_cache: torch.Size([1, 4, 263, 128]) + layer.1.v_cache: torch.Size([1, 4, 263, 128]) + layer.2.k_cache: torch.Size([1, 4, 263, 128]) + layer.2.v_cache: torch.Size([1, 4, 263, 128]) + layer.3.k_cache: torch.Size([1, 4, 263, 128]) + layer.3.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.k_cache: torch.Size([1, 4, 263, 128]) + layer.4.v_cache: torch.Size([1, 4, 263, 128]) + layer.4.output: torch.Size([1, 263, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14492303 23.17542664 + layer.0.v_cache 0.00001375 0.00492599 + layer.1.k_cache 0.50395557 3.13822026 + layer.1.v_cache 0.00000564 0.00204908 + layer.2.k_cache 0.00856250 0.44613090 + layer.2.v_cache 0.00001902 0.00550782 + layer.3.k_cache 0.02692018 1.95543598 + layer.3.v_cache 0.00002118 0.00618506 + layer.4.k_cache 0.00070790 0.11333326 + layer.4.v_cache 0.00005130 0.01225904 + layer.4.output 0.01131245 205.32663973 + ------------------------------------------------------------------------------------- + TOTAL 0.04496278 86.24387954 + (elements=2,289,152) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2289152 +Total Bytes 449320 +BPFP 1.5703 bits/point +EBPFP 3.1405 equivalent bits/point +MSE 86.243880 +---------------------- -------------------------------------------------------- +Time: 0.860s Load: 0.008s, Pack+Encode: 0.362s, Decode+Unpack: 0.490s +---------------------- -------------------------------------------------------- +💾 Converting with 86.2439 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-431.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-431.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-432.zst (82/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-432.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 276, 128) +Output shape: (1, 276, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) -> torch.Size([1, 1, 276, 512]) + layer.4.output: torch.Size([1, 276, 3584]) -> torch.Size([1, 1, 276, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,704B, BPFP=0.8324 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,240B, BPFP=2.6744 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,264B, BPFP=1.4303 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,968B, BPFP=2.6024 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,600B, BPFP=1.6757 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,464B, BPFP=2.5738 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,812B, BPFP=1.5745 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,160B, BPFP=2.6132 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,512B, BPFP=2.2369 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,568B, BPFP=2.5231 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,080B, BPFP=0.8337 +⌛️ [2/4] FRONTEND: Frontend time: 0.342s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.529s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 276, 128]) + layer.0.v_cache: torch.Size([1, 4, 276, 128]) + layer.1.k_cache: torch.Size([1, 4, 276, 128]) + layer.1.v_cache: torch.Size([1, 4, 276, 128]) + layer.2.k_cache: torch.Size([1, 4, 276, 128]) + layer.2.v_cache: torch.Size([1, 4, 276, 128]) + layer.3.k_cache: torch.Size([1, 4, 276, 128]) + layer.3.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.k_cache: torch.Size([1, 4, 276, 128]) + layer.4.v_cache: torch.Size([1, 4, 276, 128]) + layer.4.output: torch.Size([1, 276, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14113761 21.94207498 + layer.0.v_cache 0.00001369 0.00493465 + layer.1.k_cache 0.59212339 3.01653301 + layer.1.v_cache 0.00000552 0.00195251 + layer.2.k_cache 0.01129283 0.44283231 + layer.2.v_cache 0.00001853 0.00541890 + layer.3.k_cache 0.01358872 1.98513860 + layer.3.v_cache 0.00001883 0.00605158 + layer.4.k_cache 0.00070763 0.11215513 + layer.4.v_cache 0.00005153 0.01197250 + layer.4.output 0.01019159 193.57930577 + ------------------------------------------------------------------------------------- + TOTAL 0.04884114 81.32848262 + (elements=2,402,304) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2402304 +Total Bytes 469372 +BPFP 1.5631 bits/point +EBPFP 3.1261 equivalent bits/point +MSE 81.328483 +---------------------- -------------------------------------------------------- +Time: 0.880s Load: 0.009s, Pack+Encode: 0.342s, Decode+Unpack: 0.529s +---------------------- -------------------------------------------------------- +💾 Converting with 81.3285 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-432.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-432.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-434.zst (83/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-434.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 287, 128) +Output shape: (1, 287, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) -> torch.Size([1, 1, 287, 512]) + layer.4.output: torch.Size([1, 287, 3584]) -> torch.Size([1, 1, 287, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,940B, BPFP=0.8134 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 48,340B, BPFP=2.6318 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,360B, BPFP=1.4351 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,064B, BPFP=2.5623 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,672B, BPFP=1.6699 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,508B, BPFP=2.5320 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,980B, BPFP=1.5777 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,104B, BPFP=2.5645 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,792B, BPFP=2.2208 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,640B, BPFP=2.4848 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,772B, BPFP=0.7915 +⌛️ [2/4] FRONTEND: Frontend time: 0.360s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.output: torch.Size([1, 287, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.501s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 287, 128]) + layer.0.v_cache: torch.Size([1, 4, 287, 128]) + layer.1.k_cache: torch.Size([1, 4, 287, 128]) + layer.1.v_cache: torch.Size([1, 4, 287, 128]) + layer.2.k_cache: torch.Size([1, 4, 287, 128]) + layer.2.v_cache: torch.Size([1, 4, 287, 128]) + layer.3.k_cache: torch.Size([1, 4, 287, 128]) + layer.3.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.k_cache: torch.Size([1, 4, 287, 128]) + layer.4.v_cache: torch.Size([1, 4, 287, 128]) + layer.4.output: torch.Size([1, 287, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11028925 22.45168908 + layer.0.v_cache 0.00001466 0.00496726 + layer.1.k_cache 0.56032453 2.88411806 + layer.1.v_cache 0.00000595 0.00199909 + layer.2.k_cache 0.00809042 0.44987594 + layer.2.v_cache 0.00001929 0.00575089 + layer.3.k_cache 0.03501172 1.87940197 + layer.3.v_cache 0.00001861 0.00634229 + layer.4.k_cache 0.00070074 0.11615180 + layer.4.v_cache 0.00005246 0.01212126 + layer.4.output 0.00889761 188.36619587 + ------------------------------------------------------------------------------------- + TOTAL 0.04569476 79.19857581 + (elements=2,498,048) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2498048 +Total Bytes 478172 +BPFP 1.5313 bits/point +EBPFP 3.0627 equivalent bits/point +MSE 79.198576 +---------------------- -------------------------------------------------------- +Time: 0.870s Load: 0.010s, Pack+Encode: 0.360s, Decode+Unpack: 0.501s +---------------------- -------------------------------------------------------- +💾 Converting with 79.1986 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-434.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-434.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-440.zst (84/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-440.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 311, 128) +Output shape: (1, 311, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.0.v_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.1.k_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.1.v_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.2.k_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.2.v_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.3.k_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.3.v_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.4.k_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.4.v_cache: torch.Size([1, 4, 311, 128]) -> torch.Size([1, 1, 311, 512]) + layer.4.output: torch.Size([1, 311, 3584]) -> torch.Size([1, 1, 311, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,752B, BPFP=0.7914 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,864B, BPFP=2.5555 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,684B, BPFP=1.3909 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,684B, BPFP=2.4962 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,448B, BPFP=1.6302 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 49,080B, BPFP=2.4658 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,804B, BPFP=1.5476 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,896B, BPFP=2.5068 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 43,064B, BPFP=2.1636 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 48,124B, BPFP=2.4178 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 111,392B, BPFP=0.7995 +⌛️ [2/4] FRONTEND: Frontend time: 0.355s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 311, 128]) + layer.0.v_cache: torch.Size([1, 4, 311, 128]) + layer.1.k_cache: torch.Size([1, 4, 311, 128]) + layer.1.v_cache: torch.Size([1, 4, 311, 128]) + layer.2.k_cache: torch.Size([1, 4, 311, 128]) + layer.2.v_cache: torch.Size([1, 4, 311, 128]) + layer.3.k_cache: torch.Size([1, 4, 311, 128]) + layer.3.v_cache: torch.Size([1, 4, 311, 128]) + layer.4.k_cache: torch.Size([1, 4, 311, 128]) + layer.4.v_cache: torch.Size([1, 4, 311, 128]) + layer.4.output: torch.Size([1, 311, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.499s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 311, 128]) + layer.0.v_cache: torch.Size([1, 4, 311, 128]) + layer.1.k_cache: torch.Size([1, 4, 311, 128]) + layer.1.v_cache: torch.Size([1, 4, 311, 128]) + layer.2.k_cache: torch.Size([1, 4, 311, 128]) + layer.2.v_cache: torch.Size([1, 4, 311, 128]) + layer.3.k_cache: torch.Size([1, 4, 311, 128]) + layer.3.v_cache: torch.Size([1, 4, 311, 128]) + layer.4.k_cache: torch.Size([1, 4, 311, 128]) + layer.4.v_cache: torch.Size([1, 4, 311, 128]) + layer.4.output: torch.Size([1, 311, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14227538 22.46620654 + layer.0.v_cache 0.00001500 0.00503786 + layer.1.k_cache 0.64922478 3.14534885 + layer.1.v_cache 0.00000572 0.00209173 + layer.2.k_cache 0.00828172 0.44256391 + layer.2.v_cache 0.00001922 0.00547655 + layer.3.k_cache 0.02636586 1.99759804 + layer.3.v_cache 0.00001996 0.00646514 + layer.4.k_cache 0.00069494 0.11277970 + layer.4.v_cache 0.00005248 0.01261853 + layer.4.output 0.04686997 170.19843822 + ------------------------------------------------------------------------------------- + TOTAL 0.06794382 71.74030908 + (elements=2,706,944) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2706944 +Total Bytes 508792 +BPFP 1.5037 bits/point +EBPFP 3.0073 equivalent bits/point +MSE 71.740309 +---------------------- -------------------------------------------------------- +Time: 0.864s Load: 0.010s, Pack+Encode: 0.355s, Decode+Unpack: 0.499s +---------------------- -------------------------------------------------------- +💾 Converting with 71.7403 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-440.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-440.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-442.zst (85/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-442.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 255, 128) +Output shape: (1, 255, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.0.v_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.1.k_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.1.v_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.2.k_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.2.v_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.3.k_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.3.v_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.4.k_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.4.v_cache: torch.Size([1, 4, 255, 128]) -> torch.Size([1, 1, 255, 512]) + layer.4.output: torch.Size([1, 255, 3584]) -> torch.Size([1, 1, 255, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 12,740B, BPFP=0.7806 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 40,788B, BPFP=2.4993 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 22,076B, BPFP=1.3527 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 39,808B, BPFP=2.4392 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 26,016B, BPFP=1.5941 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 39,432B, BPFP=2.4162 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 24,524B, BPFP=1.5027 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 39,940B, BPFP=2.4473 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 34,576B, BPFP=2.1186 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 38,716B, BPFP=2.3723 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 99,992B, BPFP=0.8753 +⌛️ [2/4] FRONTEND: Frontend time: 0.369s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 255, 128]) + layer.0.v_cache: torch.Size([1, 4, 255, 128]) + layer.1.k_cache: torch.Size([1, 4, 255, 128]) + layer.1.v_cache: torch.Size([1, 4, 255, 128]) + layer.2.k_cache: torch.Size([1, 4, 255, 128]) + layer.2.v_cache: torch.Size([1, 4, 255, 128]) + layer.3.k_cache: torch.Size([1, 4, 255, 128]) + layer.3.v_cache: torch.Size([1, 4, 255, 128]) + layer.4.k_cache: torch.Size([1, 4, 255, 128]) + layer.4.v_cache: torch.Size([1, 4, 255, 128]) + layer.4.output: torch.Size([1, 255, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.402s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 255, 128]) + layer.0.v_cache: torch.Size([1, 4, 255, 128]) + layer.1.k_cache: torch.Size([1, 4, 255, 128]) + layer.1.v_cache: torch.Size([1, 4, 255, 128]) + layer.2.k_cache: torch.Size([1, 4, 255, 128]) + layer.2.v_cache: torch.Size([1, 4, 255, 128]) + layer.3.k_cache: torch.Size([1, 4, 255, 128]) + layer.3.v_cache: torch.Size([1, 4, 255, 128]) + layer.4.k_cache: torch.Size([1, 4, 255, 128]) + layer.4.v_cache: torch.Size([1, 4, 255, 128]) + layer.4.output: torch.Size([1, 255, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13149661 22.03169233 + layer.0.v_cache 0.00001370 0.00487628 + layer.1.k_cache 0.50214604 2.74353339 + layer.1.v_cache 0.00000625 0.00214069 + layer.2.k_cache 0.00807812 0.42828839 + layer.2.v_cache 0.00001969 0.00556415 + layer.3.k_cache 0.06061829 1.94557591 + layer.3.v_cache 0.00001868 0.00620918 + layer.4.k_cache 0.00068906 0.11340305 + layer.4.v_cache 0.00005044 0.01231992 + layer.4.output 1.20679682 173.32167367 + ------------------------------------------------------------------------------------- + TOTAL 0.53827733 72.97325406 + (elements=2,219,520) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2219520 +Total Bytes 418608 +BPFP 1.5088 bits/point +EBPFP 3.0176 equivalent bits/point +MSE 72.973254 +---------------------- -------------------------------------------------------- +Time: 0.780s Load: 0.009s, Pack+Encode: 0.369s, Decode+Unpack: 0.402s +---------------------- -------------------------------------------------------- +💾 Converting with 72.9733 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-442.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-442.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-453.zst (86/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-453.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 300, 128) +Output shape: (1, 300, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.0.v_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.1.k_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.1.v_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.2.k_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.2.v_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.3.k_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.3.v_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.4.k_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.4.v_cache: torch.Size([1, 4, 300, 128]) -> torch.Size([1, 1, 300, 512]) + layer.4.output: torch.Size([1, 300, 3584]) -> torch.Size([1, 1, 300, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,876B, BPFP=0.8269 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,208B, BPFP=2.6150 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,376B, BPFP=1.4258 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,948B, BPFP=2.5494 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,872B, BPFP=1.6600 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,400B, BPFP=2.5208 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,980B, BPFP=1.5615 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,100B, BPFP=2.5573 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,264B, BPFP=2.2012 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,324B, BPFP=2.4648 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,040B, BPFP=0.7518 +⌛️ [2/4] FRONTEND: Frontend time: 0.318s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 300, 128]) + layer.0.v_cache: torch.Size([1, 4, 300, 128]) + layer.1.k_cache: torch.Size([1, 4, 300, 128]) + layer.1.v_cache: torch.Size([1, 4, 300, 128]) + layer.2.k_cache: torch.Size([1, 4, 300, 128]) + layer.2.v_cache: torch.Size([1, 4, 300, 128]) + layer.3.k_cache: torch.Size([1, 4, 300, 128]) + layer.3.v_cache: torch.Size([1, 4, 300, 128]) + layer.4.k_cache: torch.Size([1, 4, 300, 128]) + layer.4.v_cache: torch.Size([1, 4, 300, 128]) + layer.4.output: torch.Size([1, 300, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.484s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 300, 128]) + layer.0.v_cache: torch.Size([1, 4, 300, 128]) + layer.1.k_cache: torch.Size([1, 4, 300, 128]) + layer.1.v_cache: torch.Size([1, 4, 300, 128]) + layer.2.k_cache: torch.Size([1, 4, 300, 128]) + layer.2.v_cache: torch.Size([1, 4, 300, 128]) + layer.3.k_cache: torch.Size([1, 4, 300, 128]) + layer.3.v_cache: torch.Size([1, 4, 300, 128]) + layer.4.k_cache: torch.Size([1, 4, 300, 128]) + layer.4.v_cache: torch.Size([1, 4, 300, 128]) + layer.4.output: torch.Size([1, 300, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13491969 22.36704915 + layer.0.v_cache 0.00001558 0.00483397 + layer.1.k_cache 0.64590566 3.06443990 + layer.1.v_cache 0.00000556 0.00193051 + layer.2.k_cache 0.01395306 0.45390162 + layer.2.v_cache 0.00001972 0.00541022 + layer.3.k_cache 0.02379198 2.02786214 + layer.3.v_cache 0.00001942 0.00626970 + layer.4.k_cache 0.00073123 0.11514879 + layer.4.v_cache 0.00005274 0.01218880 + layer.4.output 0.05011183 176.41779762 + ------------------------------------------------------------------------------------- + TOTAL 0.06883514 74.29315401 + (elements=2,611,200) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2611200 +Total Bytes 492388 +BPFP 1.5085 bits/point +EBPFP 3.0171 equivalent bits/point +MSE 74.293154 +---------------------- -------------------------------------------------------- +Time: 0.810s Load: 0.009s, Pack+Encode: 0.318s, Decode+Unpack: 0.484s +---------------------- -------------------------------------------------------- +💾 Converting with 74.2932 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-453.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-453.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-47.zst (87/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-47.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 291, 128) +Output shape: (1, 291, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) -> torch.Size([1, 1, 291, 512]) + layer.4.output: torch.Size([1, 291, 3584]) -> torch.Size([1, 1, 291, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,016B, BPFP=0.8063 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,136B, BPFP=2.6383 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,544B, BPFP=1.4253 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 47,908B, BPFP=2.5724 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,872B, BPFP=1.6576 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,024B, BPFP=2.5249 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,208B, BPFP=1.5683 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,664B, BPFP=2.5593 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,176B, BPFP=2.2109 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,108B, BPFP=2.4757 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 100,880B, BPFP=0.7738 +⌛️ [2/4] FRONTEND: Frontend time: 0.316s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.output: torch.Size([1, 291, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.486s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 291, 128]) + layer.0.v_cache: torch.Size([1, 4, 291, 128]) + layer.1.k_cache: torch.Size([1, 4, 291, 128]) + layer.1.v_cache: torch.Size([1, 4, 291, 128]) + layer.2.k_cache: torch.Size([1, 4, 291, 128]) + layer.2.v_cache: torch.Size([1, 4, 291, 128]) + layer.3.k_cache: torch.Size([1, 4, 291, 128]) + layer.3.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.k_cache: torch.Size([1, 4, 291, 128]) + layer.4.v_cache: torch.Size([1, 4, 291, 128]) + layer.4.output: torch.Size([1, 291, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14826349 22.16207078 + layer.0.v_cache 0.00001415 0.00493233 + layer.1.k_cache 0.61440683 3.06437216 + layer.1.v_cache 0.00000562 0.00213495 + layer.2.k_cache 0.00551110 0.42808805 + layer.2.v_cache 0.00001912 0.00565015 + layer.3.k_cache 0.01294636 1.94373356 + layer.3.v_cache 0.00002032 0.00632630 + layer.4.k_cache 0.00070633 0.11641392 + layer.4.v_cache 0.00005074 0.01215060 + layer.4.output 0.04972342 179.83615611 + ------------------------------------------------------------------------------------- + TOTAL 0.06647106 75.68229209 + (elements=2,532,864) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2532864 +Total Bytes 481536 +BPFP 1.5209 bits/point +EBPFP 3.0418 equivalent bits/point +MSE 75.682292 +---------------------- -------------------------------------------------------- +Time: 0.813s Load: 0.010s, Pack+Encode: 0.316s, Decode+Unpack: 0.486s +---------------------- -------------------------------------------------------- +💾 Converting with 75.6823 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-47.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-47.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-5.zst (88/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-5.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 277, 128) +Output shape: (1, 277, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) -> torch.Size([1, 1, 277, 512]) + layer.4.output: torch.Size([1, 277, 3584]) -> torch.Size([1, 1, 277, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,480B, BPFP=0.8168 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,144B, BPFP=2.6593 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,316B, BPFP=1.4280 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,924B, BPFP=2.5905 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,412B, BPFP=1.6591 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,340B, BPFP=2.5575 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,844B, BPFP=1.5706 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,092B, BPFP=2.6000 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,468B, BPFP=2.2263 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,412B, BPFP=2.5052 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 98,616B, BPFP=0.7947 +⌛️ [2/4] FRONTEND: Frontend time: 0.334s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.468s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 277, 128]) + layer.0.v_cache: torch.Size([1, 4, 277, 128]) + layer.1.k_cache: torch.Size([1, 4, 277, 128]) + layer.1.v_cache: torch.Size([1, 4, 277, 128]) + layer.2.k_cache: torch.Size([1, 4, 277, 128]) + layer.2.v_cache: torch.Size([1, 4, 277, 128]) + layer.3.k_cache: torch.Size([1, 4, 277, 128]) + layer.3.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.k_cache: torch.Size([1, 4, 277, 128]) + layer.4.v_cache: torch.Size([1, 4, 277, 128]) + layer.4.output: torch.Size([1, 277, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13861307 21.94341049 + layer.0.v_cache 0.00001419 0.00499906 + layer.1.k_cache 0.53867822 3.00535765 + layer.1.v_cache 0.00000573 0.00203128 + layer.2.k_cache 0.00871976 0.43135286 + layer.2.v_cache 0.00001877 0.00522013 + layer.3.k_cache 0.02195022 1.90467498 + layer.3.v_cache 0.00001967 0.00626011 + layer.4.k_cache 0.00069227 0.11289477 + layer.4.v_cache 0.00004961 0.01132106 + layer.4.output 0.00954030 197.71731563 + ------------------------------------------------------------------------------------- + TOTAL 0.04562021 83.02639599 + (elements=2,411,008) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2411008 +Total Bytes 464048 +BPFP 1.5398 bits/point +EBPFP 3.0795 equivalent bits/point +MSE 83.026396 +---------------------- -------------------------------------------------------- +Time: 0.811s Load: 0.009s, Pack+Encode: 0.334s, Decode+Unpack: 0.468s +---------------------- -------------------------------------------------------- +💾 Converting with 83.0264 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-5.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-5.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-61.zst (89/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-61.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 286, 128) +Output shape: (1, 286, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) -> torch.Size([1, 1, 286, 512]) + layer.4.output: torch.Size([1, 286, 3584]) -> torch.Size([1, 1, 286, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,264B, BPFP=0.8339 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,980B, BPFP=2.6213 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,272B, BPFP=1.4353 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,780B, BPFP=2.5557 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,612B, BPFP=1.6724 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,344B, BPFP=2.5319 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,952B, BPFP=1.5817 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 47,032B, BPFP=2.5695 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,664B, BPFP=2.2216 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,444B, BPFP=2.4827 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,448B, BPFP=0.8074 +⌛️ [2/4] FRONTEND: Frontend time: 0.382s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.497s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 286, 128]) + layer.0.v_cache: torch.Size([1, 4, 286, 128]) + layer.1.k_cache: torch.Size([1, 4, 286, 128]) + layer.1.v_cache: torch.Size([1, 4, 286, 128]) + layer.2.k_cache: torch.Size([1, 4, 286, 128]) + layer.2.v_cache: torch.Size([1, 4, 286, 128]) + layer.3.k_cache: torch.Size([1, 4, 286, 128]) + layer.3.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.k_cache: torch.Size([1, 4, 286, 128]) + layer.4.v_cache: torch.Size([1, 4, 286, 128]) + layer.4.output: torch.Size([1, 286, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13928204 22.92604929 + layer.0.v_cache 0.00001373 0.00484985 + layer.1.k_cache 0.60860016 3.00362327 + layer.1.v_cache 0.00000555 0.00201384 + layer.2.k_cache 0.01026906 0.45295198 + layer.2.v_cache 0.00001872 0.00554439 + layer.3.k_cache 0.05166252 2.01043829 + layer.3.v_cache 0.00001937 0.00640236 + layer.4.k_cache 0.00073448 0.11658040 + layer.4.v_cache 0.00005026 0.01211357 + layer.4.output 0.00697322 192.37434441 + ------------------------------------------------------------------------------------- + TOTAL 0.05055697 80.89182224 + (elements=2,489,344) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2489344 +Total Bytes 478792 +BPFP 1.5387 bits/point +EBPFP 3.0774 equivalent bits/point +MSE 80.891822 +---------------------- -------------------------------------------------------- +Time: 0.889s Load: 0.010s, Pack+Encode: 0.382s, Decode+Unpack: 0.497s +---------------------- -------------------------------------------------------- +💾 Converting with 80.8918 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-61.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-61.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-62.zst (90/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-62.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 295, 128) +Output shape: (1, 295, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) -> torch.Size([1, 1, 295, 512]) + layer.4.output: torch.Size([1, 295, 3584]) -> torch.Size([1, 1, 295, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,320B, BPFP=0.8114 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,644B, BPFP=2.6294 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,904B, BPFP=1.4250 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,392B, BPFP=2.5631 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,304B, BPFP=1.6581 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,708B, BPFP=2.5269 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,512B, BPFP=1.5631 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,312B, BPFP=2.5589 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,896B, BPFP=2.2191 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,832B, BPFP=2.4805 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 105,108B, BPFP=0.7953 +⌛️ [2/4] FRONTEND: Frontend time: 0.344s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.output: torch.Size([1, 295, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.478s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 295, 128]) + layer.0.v_cache: torch.Size([1, 4, 295, 128]) + layer.1.k_cache: torch.Size([1, 4, 295, 128]) + layer.1.v_cache: torch.Size([1, 4, 295, 128]) + layer.2.k_cache: torch.Size([1, 4, 295, 128]) + layer.2.v_cache: torch.Size([1, 4, 295, 128]) + layer.3.k_cache: torch.Size([1, 4, 295, 128]) + layer.3.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.k_cache: torch.Size([1, 4, 295, 128]) + layer.4.v_cache: torch.Size([1, 4, 295, 128]) + layer.4.output: torch.Size([1, 295, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13820392 21.98727986 + layer.0.v_cache 0.00001454 0.00493266 + layer.1.k_cache 0.61684219 3.18620688 + layer.1.v_cache 0.00000570 0.00201515 + layer.2.k_cache 0.01285210 0.43581915 + layer.2.v_cache 0.00002011 0.00554291 + layer.3.k_cache 0.05980752 2.06297111 + layer.3.v_cache 0.00002263 0.00633426 + layer.4.k_cache 0.00071023 0.11512237 + layer.4.v_cache 0.00005833 0.01262332 + layer.4.output 0.05086143 178.84164649 + ------------------------------------------------------------------------------------- + TOTAL 0.06968043 75.27708077 + (elements=2,567,680) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2567680 +Total Bytes 490932 +BPFP 1.5296 bits/point +EBPFP 3.0591 equivalent bits/point +MSE 75.277081 +---------------------- -------------------------------------------------------- +Time: 0.830s Load: 0.009s, Pack+Encode: 0.344s, Decode+Unpack: 0.478s +---------------------- -------------------------------------------------------- +💾 Converting with 75.2771 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-62.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-62.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-63.zst (91/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-63.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 301, 128) +Output shape: (1, 301, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) -> torch.Size([1, 1, 301, 512]) + layer.4.output: torch.Size([1, 301, 3584]) -> torch.Size([1, 1, 301, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,460B, BPFP=0.8025 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,408B, BPFP=2.6167 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,516B, BPFP=1.4284 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,148B, BPFP=2.5513 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,860B, BPFP=1.6539 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 48,572B, BPFP=2.5214 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,188B, BPFP=1.5671 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,212B, BPFP=2.5546 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,468B, BPFP=2.2045 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 47,616B, BPFP=2.4718 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 106,976B, BPFP=0.7933 +⌛️ [2/4] FRONTEND: Frontend time: 0.332s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.output: torch.Size([1, 301, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.469s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 301, 128]) + layer.0.v_cache: torch.Size([1, 4, 301, 128]) + layer.1.k_cache: torch.Size([1, 4, 301, 128]) + layer.1.v_cache: torch.Size([1, 4, 301, 128]) + layer.2.k_cache: torch.Size([1, 4, 301, 128]) + layer.2.v_cache: torch.Size([1, 4, 301, 128]) + layer.3.k_cache: torch.Size([1, 4, 301, 128]) + layer.3.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.k_cache: torch.Size([1, 4, 301, 128]) + layer.4.v_cache: torch.Size([1, 4, 301, 128]) + layer.4.output: torch.Size([1, 301, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14825445 22.37797349 + layer.0.v_cache 0.00001396 0.00525452 + layer.1.k_cache 0.63210902 3.06583483 + layer.1.v_cache 0.00000581 0.00221694 + layer.2.k_cache 0.01184132 0.44293177 + layer.2.v_cache 0.00001890 0.00598288 + layer.3.k_cache 0.01767834 1.94057163 + layer.3.v_cache 0.00002009 0.00675281 + layer.4.k_cache 0.00068272 0.12099050 + layer.4.v_cache 0.00005247 0.01292484 + layer.4.output 0.04867053 178.38455150 + ------------------------------------------------------------------------------------- + TOTAL 0.06772769 75.09842910 + (elements=2,619,904) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2619904 +Total Bytes 499424 +BPFP 1.5250 bits/point +EBPFP 3.0500 equivalent bits/point +MSE 75.098429 +---------------------- -------------------------------------------------------- +Time: 0.811s Load: 0.010s, Pack+Encode: 0.332s, Decode+Unpack: 0.469s +---------------------- -------------------------------------------------------- +💾 Converting with 75.0984 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-63.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-63.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-66.zst (92/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-66.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 264, 128) +Output shape: (1, 264, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.0.v_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.1.k_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.1.v_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.2.k_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.2.v_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.3.k_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.3.v_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.4.k_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.4.v_cache: torch.Size([1, 4, 264, 128]) -> torch.Size([1, 1, 264, 512]) + layer.4.output: torch.Size([1, 264, 3584]) -> torch.Size([1, 1, 264, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,004B, BPFP=0.8288 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,136B, BPFP=2.7306 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,448B, BPFP=1.4470 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,024B, BPFP=2.6648 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,692B, BPFP=1.6982 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,332B, BPFP=2.6238 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 26,824B, BPFP=1.5876 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,064B, BPFP=2.6671 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 38,504B, BPFP=2.2789 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 43,440B, BPFP=2.5710 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 91,296B, BPFP=0.7719 +⌛️ [2/4] FRONTEND: Frontend time: 0.328s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 264, 128]) + layer.0.v_cache: torch.Size([1, 4, 264, 128]) + layer.1.k_cache: torch.Size([1, 4, 264, 128]) + layer.1.v_cache: torch.Size([1, 4, 264, 128]) + layer.2.k_cache: torch.Size([1, 4, 264, 128]) + layer.2.v_cache: torch.Size([1, 4, 264, 128]) + layer.3.k_cache: torch.Size([1, 4, 264, 128]) + layer.3.v_cache: torch.Size([1, 4, 264, 128]) + layer.4.k_cache: torch.Size([1, 4, 264, 128]) + layer.4.v_cache: torch.Size([1, 4, 264, 128]) + layer.4.output: torch.Size([1, 264, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.494s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 264, 128]) + layer.0.v_cache: torch.Size([1, 4, 264, 128]) + layer.1.k_cache: torch.Size([1, 4, 264, 128]) + layer.1.v_cache: torch.Size([1, 4, 264, 128]) + layer.2.k_cache: torch.Size([1, 4, 264, 128]) + layer.2.v_cache: torch.Size([1, 4, 264, 128]) + layer.3.k_cache: torch.Size([1, 4, 264, 128]) + layer.3.v_cache: torch.Size([1, 4, 264, 128]) + layer.4.k_cache: torch.Size([1, 4, 264, 128]) + layer.4.v_cache: torch.Size([1, 4, 264, 128]) + layer.4.output: torch.Size([1, 264, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14315909 22.59406072 + layer.0.v_cache 0.00001391 0.00507402 + layer.1.k_cache 0.49711875 3.27705499 + layer.1.v_cache 0.00000560 0.00209790 + layer.2.k_cache 0.00646762 0.44222716 + layer.2.v_cache 0.00001905 0.00570502 + layer.3.k_cache 0.04707547 2.08054976 + layer.3.v_cache 0.00001924 0.00649147 + layer.4.k_cache 0.00071833 0.11526116 + layer.4.v_cache 0.00005197 0.01206611 + layer.4.output 0.00997522 205.89280641 + ------------------------------------------------------------------------------------- + TOTAL 0.04496915 86.45824901 + (elements=2,297,856) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2297856 +Total Bytes 447764 +BPFP 1.5589 bits/point +EBPFP 3.1178 equivalent bits/point +MSE 86.458249 +---------------------- -------------------------------------------------------- +Time: 0.831s Load: 0.008s, Pack+Encode: 0.328s, Decode+Unpack: 0.494s +---------------------- -------------------------------------------------------- +💾 Converting with 86.4582 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-66.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-66.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-75.zst (93/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-75.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.008s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 279, 128) +Output shape: (1, 279, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) -> torch.Size([1, 1, 279, 512]) + layer.4.output: torch.Size([1, 279, 3584]) -> torch.Size([1, 1, 279, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,692B, BPFP=0.8228 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,300B, BPFP=2.6490 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 25,336B, BPFP=1.4189 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,992B, BPFP=2.5757 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,908B, BPFP=1.6750 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,572B, BPFP=2.5522 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,972B, BPFP=1.5665 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,136B, BPFP=2.5838 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,692B, BPFP=2.2229 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,692B, BPFP=2.5029 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 96,924B, BPFP=0.7754 +⌛️ [2/4] FRONTEND: Frontend time: 0.350s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.480s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 279, 128]) + layer.0.v_cache: torch.Size([1, 4, 279, 128]) + layer.1.k_cache: torch.Size([1, 4, 279, 128]) + layer.1.v_cache: torch.Size([1, 4, 279, 128]) + layer.2.k_cache: torch.Size([1, 4, 279, 128]) + layer.2.v_cache: torch.Size([1, 4, 279, 128]) + layer.3.k_cache: torch.Size([1, 4, 279, 128]) + layer.3.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.k_cache: torch.Size([1, 4, 279, 128]) + layer.4.v_cache: torch.Size([1, 4, 279, 128]) + layer.4.output: torch.Size([1, 279, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15581964 22.82425900 + layer.0.v_cache 0.00001703 0.00492896 + layer.1.k_cache 0.56298232 2.98486547 + layer.1.v_cache 0.00000550 0.00202015 + layer.2.k_cache 0.00979712 0.42011685 + layer.2.v_cache 0.00002023 0.00574336 + layer.3.k_cache 0.04960944 1.94417493 + layer.3.v_cache 0.00001971 0.00630456 + layer.4.k_cache 0.00068561 0.11193868 + layer.4.v_cache 0.00005245 0.01204676 + layer.4.output 0.01122868 195.36111111 + ------------------------------------------------------------------------------------- + TOTAL 0.05044764 82.10848097 + (elements=2,428,416) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2428416 +Total Bytes 464216 +BPFP 1.5293 bits/point +EBPFP 3.0586 equivalent bits/point +MSE 82.108481 +---------------------- -------------------------------------------------------- +Time: 0.838s Load: 0.008s, Pack+Encode: 0.350s, Decode+Unpack: 0.480s +---------------------- -------------------------------------------------------- +💾 Converting with 82.1085 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-75.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-75.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-76.zst (94/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-76.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,888B, BPFP=0.8220 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,776B, BPFP=2.6378 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,068B, BPFP=1.4393 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,456B, BPFP=2.5649 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,352B, BPFP=1.6758 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,916B, BPFP=2.5351 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,460B, BPFP=1.5713 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,468B, BPFP=2.5656 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,244B, BPFP=2.2220 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,008B, BPFP=2.4850 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 102,828B, BPFP=0.8110 +⌛️ [2/4] FRONTEND: Frontend time: 0.358s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.572s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13599485 22.46663124 + layer.0.v_cache 0.00001418 0.00485671 + layer.1.k_cache 0.56373192 3.11615033 + layer.1.v_cache 0.00000588 0.00208605 + layer.2.k_cache 0.01549433 0.43137521 + layer.2.v_cache 0.00001867 0.00537703 + layer.3.k_cache 0.03475572 1.88185837 + layer.3.v_cache 0.00001907 0.00607677 + layer.4.k_cache 0.00072111 0.11613967 + layer.4.v_cache 0.00005295 0.01178376 + layer.4.output 0.00820773 190.88709932 + ------------------------------------------------------------------------------------- + TOTAL 0.04754487 80.25011943 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 474464 +BPFP 1.5409 bits/point +EBPFP 3.0819 equivalent bits/point +MSE 80.250119 +---------------------- -------------------------------------------------------- +Time: 0.940s Load: 0.009s, Pack+Encode: 0.358s, Decode+Unpack: 0.572s +---------------------- -------------------------------------------------------- +💾 Converting with 80.2501 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-76.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-76.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-79.zst (95/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-79.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.010s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 308, 128) +Output shape: (1, 308, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) -> torch.Size([1, 1, 308, 512]) + layer.4.output: torch.Size([1, 308, 3584]) -> torch.Size([1, 1, 308, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,676B, BPFP=0.7953 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 50,648B, BPFP=2.5694 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 27,724B, BPFP=1.4065 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 49,448B, BPFP=2.5085 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 32,400B, BPFP=1.6437 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 49,048B, BPFP=2.4882 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 30,636B, BPFP=1.5542 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 49,748B, BPFP=2.5237 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 42,976B, BPFP=2.1802 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 48,152B, BPFP=2.4428 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,348B, BPFP=0.7490 +⌛️ [2/4] FRONTEND: Frontend time: 0.363s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.output: torch.Size([1, 308, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.519s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 308, 128]) + layer.0.v_cache: torch.Size([1, 4, 308, 128]) + layer.1.k_cache: torch.Size([1, 4, 308, 128]) + layer.1.v_cache: torch.Size([1, 4, 308, 128]) + layer.2.k_cache: torch.Size([1, 4, 308, 128]) + layer.2.v_cache: torch.Size([1, 4, 308, 128]) + layer.3.k_cache: torch.Size([1, 4, 308, 128]) + layer.3.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.k_cache: torch.Size([1, 4, 308, 128]) + layer.4.v_cache: torch.Size([1, 4, 308, 128]) + layer.4.output: torch.Size([1, 308, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13331154 22.54979518 + layer.0.v_cache 0.00001460 0.00483904 + layer.1.k_cache 0.64535681 2.97157823 + layer.1.v_cache 0.00000563 0.00213104 + layer.2.k_cache 0.00762887 0.41579799 + layer.2.v_cache 0.00001910 0.00575880 + layer.3.k_cache 0.03538683 1.82370570 + layer.3.v_cache 0.00002103 0.00658057 + layer.4.k_cache 0.00072257 0.11636888 + layer.4.v_cache 0.00005300 0.01244254 + layer.4.output 0.04522907 169.51555253 + ------------------------------------------------------------------------------------- + TOTAL 0.06700726 71.44222739 + (elements=2,680,832) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2680832 +Total Bytes 499804 +BPFP 1.4915 bits/point +EBPFP 2.9830 equivalent bits/point +MSE 71.442227 +---------------------- -------------------------------------------------------- +Time: 0.892s Load: 0.010s, Pack+Encode: 0.363s, Decode+Unpack: 0.519s +---------------------- -------------------------------------------------------- +💾 Converting with 71.4422 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-79.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-79.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-83.zst (96/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-83.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 284, 128) +Output shape: (1, 284, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) -> torch.Size([1, 1, 284, 512]) + layer.4.output: torch.Size([1, 284, 3584]) -> torch.Size([1, 1, 284, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,980B, BPFP=0.8242 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,936B, BPFP=2.6373 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,084B, BPFP=1.4351 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,744B, BPFP=2.5717 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,500B, BPFP=1.6780 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,120B, BPFP=2.5374 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,520B, BPFP=1.5691 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,732B, BPFP=2.5711 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,440B, BPFP=2.2249 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,292B, BPFP=2.4919 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 103,208B, BPFP=0.8112 +⌛️ [2/4] FRONTEND: Frontend time: 0.356s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.output: torch.Size([1, 284, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.534s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 284, 128]) + layer.0.v_cache: torch.Size([1, 4, 284, 128]) + layer.1.k_cache: torch.Size([1, 4, 284, 128]) + layer.1.v_cache: torch.Size([1, 4, 284, 128]) + layer.2.k_cache: torch.Size([1, 4, 284, 128]) + layer.2.v_cache: torch.Size([1, 4, 284, 128]) + layer.3.k_cache: torch.Size([1, 4, 284, 128]) + layer.3.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.k_cache: torch.Size([1, 4, 284, 128]) + layer.4.v_cache: torch.Size([1, 4, 284, 128]) + layer.4.output: torch.Size([1, 284, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14806804 22.53323751 + layer.0.v_cache 0.00001376 0.00492580 + layer.1.k_cache 0.60662186 3.11420838 + layer.1.v_cache 0.00000571 0.00203189 + layer.2.k_cache 0.01239752 0.43407693 + layer.2.v_cache 0.00002027 0.00555266 + layer.3.k_cache 0.02456893 2.00051364 + layer.3.v_cache 0.00001975 0.00618216 + layer.4.k_cache 0.00069718 0.11513476 + layer.4.v_cache 0.00005379 0.01224925 + layer.4.output 0.00818182 190.58695926 + ------------------------------------------------------------------------------------- + TOTAL 0.04998468 80.13746046 + (elements=2,471,936) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2471936 +Total Bytes 476556 +BPFP 1.5423 bits/point +EBPFP 3.0846 equivalent bits/point +MSE 80.137460 +---------------------- -------------------------------------------------------- +Time: 0.901s Load: 0.011s, Pack+Encode: 0.356s, Decode+Unpack: 0.534s +---------------------- -------------------------------------------------------- +💾 Converting with 80.1375 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-83.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-83.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-87.zst (97/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-87.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 283, 128) +Output shape: (1, 283, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) -> torch.Size([1, 1, 283, 512]) + layer.4.output: torch.Size([1, 283, 3584]) -> torch.Size([1, 1, 283, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,920B, BPFP=0.8238 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 47,908B, BPFP=2.6451 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,216B, BPFP=1.4474 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 46,652B, BPFP=2.5758 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 30,588B, BPFP=1.6888 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 46,068B, BPFP=2.5435 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 28,576B, BPFP=1.5777 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 46,708B, BPFP=2.5788 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 40,504B, BPFP=2.2363 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 45,268B, BPFP=2.4993 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 101,880B, BPFP=0.8036 +⌛️ [2/4] FRONTEND: Frontend time: 0.368s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.502s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 283, 128]) + layer.0.v_cache: torch.Size([1, 4, 283, 128]) + layer.1.k_cache: torch.Size([1, 4, 283, 128]) + layer.1.v_cache: torch.Size([1, 4, 283, 128]) + layer.2.k_cache: torch.Size([1, 4, 283, 128]) + layer.2.v_cache: torch.Size([1, 4, 283, 128]) + layer.3.k_cache: torch.Size([1, 4, 283, 128]) + layer.3.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.k_cache: torch.Size([1, 4, 283, 128]) + layer.4.v_cache: torch.Size([1, 4, 283, 128]) + layer.4.output: torch.Size([1, 283, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14592592 21.97763396 + layer.0.v_cache 0.00001391 0.00492877 + layer.1.k_cache 0.53254910 2.91184237 + layer.1.v_cache 0.00000583 0.00213314 + layer.2.k_cache 0.01029040 0.41322480 + layer.2.v_cache 0.00002009 0.00573519 + layer.3.k_cache 0.03848921 2.10221922 + layer.3.v_cache 0.00001972 0.00639689 + layer.4.k_cache 0.00069417 0.11710956 + layer.4.v_cache 0.00005427 0.01261991 + layer.4.output 0.01004909 190.78243311 + ------------------------------------------------------------------------------------- + TOTAL 0.04696507 80.17828680 + (elements=2,463,232) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2463232 +Total Bytes 475288 +BPFP 1.5436 bits/point +EBPFP 3.0872 equivalent bits/point +MSE 80.178287 +---------------------- -------------------------------------------------------- +Time: 0.879s Load: 0.009s, Pack+Encode: 0.368s, Decode+Unpack: 0.502s +---------------------- -------------------------------------------------------- +💾 Converting with 80.1783 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-87.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-87.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-94.zst (98/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-94.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 272, 128) +Output shape: (1, 272, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) -> torch.Size([1, 1, 272, 512]) + layer.4.output: torch.Size([1, 272, 3584]) -> torch.Size([1, 1, 272, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,164B, BPFP=0.8136 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,812B, BPFP=2.6891 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,992B, BPFP=1.4357 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,620B, BPFP=2.6206 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 29,152B, BPFP=1.6746 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 45,068B, BPFP=2.5889 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,508B, BPFP=1.5802 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,588B, BPFP=2.6188 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,216B, BPFP=2.2528 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,144B, BPFP=2.5358 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 94,168B, BPFP=0.7728 +⌛️ [2/4] FRONTEND: Frontend time: 0.362s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.479s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 272, 128]) + layer.0.v_cache: torch.Size([1, 4, 272, 128]) + layer.1.k_cache: torch.Size([1, 4, 272, 128]) + layer.1.v_cache: torch.Size([1, 4, 272, 128]) + layer.2.k_cache: torch.Size([1, 4, 272, 128]) + layer.2.v_cache: torch.Size([1, 4, 272, 128]) + layer.3.k_cache: torch.Size([1, 4, 272, 128]) + layer.3.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.k_cache: torch.Size([1, 4, 272, 128]) + layer.4.v_cache: torch.Size([1, 4, 272, 128]) + layer.4.output: torch.Size([1, 272, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15384994 22.57724717 + layer.0.v_cache 0.00001404 0.00507525 + layer.1.k_cache 0.48768111 3.09867141 + layer.1.v_cache 0.00000584 0.00216126 + layer.2.k_cache 0.00538594 0.39707302 + layer.2.v_cache 0.00001948 0.00554554 + layer.3.k_cache 0.02347840 1.97001581 + layer.3.v_cache 0.00001899 0.00618939 + layer.4.k_cache 0.00068094 0.11362349 + layer.4.v_cache 0.00005116 0.01204588 + layer.4.output 0.00909345 195.97101497 + ------------------------------------------------------------------------------------- + TOTAL 0.04322588 82.35204429 + (elements=2,367,488) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2367488 +Total Bytes 456432 +BPFP 1.5423 bits/point +EBPFP 3.0847 equivalent bits/point +MSE 82.352044 +---------------------- -------------------------------------------------------- +Time: 0.850s Load: 0.009s, Pack+Encode: 0.362s, Decode+Unpack: 0.479s +---------------------- -------------------------------------------------------- +💾 Converting with 82.3520 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-94.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-94.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-97.zst (99/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-97.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.011s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 294, 128) +Output shape: (1, 294, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) -> torch.Size([1, 1, 294, 512]) + layer.4.output: torch.Size([1, 294, 3584]) -> torch.Size([1, 1, 294, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 15,428B, BPFP=0.8199 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 49,620B, BPFP=2.6371 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 26,964B, BPFP=1.4330 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 48,556B, BPFP=2.5806 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 31,460B, BPFP=1.6720 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 47,824B, BPFP=2.5417 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 29,788B, BPFP=1.5831 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 48,432B, BPFP=2.5740 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 41,796B, BPFP=2.2213 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 46,640B, BPFP=2.4787 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 97,388B, BPFP=0.7394 +⌛️ [2/4] FRONTEND: Frontend time: 0.311s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.477s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 294, 128]) + layer.0.v_cache: torch.Size([1, 4, 294, 128]) + layer.1.k_cache: torch.Size([1, 4, 294, 128]) + layer.1.v_cache: torch.Size([1, 4, 294, 128]) + layer.2.k_cache: torch.Size([1, 4, 294, 128]) + layer.2.v_cache: torch.Size([1, 4, 294, 128]) + layer.3.k_cache: torch.Size([1, 4, 294, 128]) + layer.3.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.k_cache: torch.Size([1, 4, 294, 128]) + layer.4.v_cache: torch.Size([1, 4, 294, 128]) + layer.4.output: torch.Size([1, 294, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13453574 22.66520847 + layer.0.v_cache 0.00001386 0.00489416 + layer.1.k_cache 0.70462929 3.16157999 + layer.1.v_cache 0.00000580 0.00210630 + layer.2.k_cache 0.00896267 0.43558025 + layer.2.v_cache 0.00001881 0.00559692 + layer.3.k_cache 0.02468209 1.89797579 + layer.3.v_cache 0.00001940 0.00632428 + layer.4.k_cache 0.00068809 0.11540583 + layer.4.v_cache 0.00005209 0.01208161 + layer.4.output 0.04953035 182.57423773 + ------------------------------------------------------------------------------------- + TOTAL 0.07178355 76.84273045 + (elements=2,558,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2558976 +Total Bytes 483896 +BPFP 1.5128 bits/point +EBPFP 3.0256 equivalent bits/point +MSE 76.842730 +---------------------- -------------------------------------------------------- +Time: 0.799s Load: 0.011s, Pack+Encode: 0.311s, Decode+Unpack: 0.477s +---------------------- -------------------------------------------------------- +💾 Converting with 76.8427 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-97.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-97.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-99.zst (100/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-99.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.009s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 268, 128) +Output shape: (1, 268, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) -> torch.Size([1, 1, 268, 512]) + layer.4.output: torch.Size([1, 268, 3584]) -> torch.Size([1, 1, 268, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 14,132B, BPFP=0.8239 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 46,764B, BPFP=2.7264 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 24,696B, BPFP=1.4398 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 45,424B, BPFP=2.6483 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 28,928B, BPFP=1.6866 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 44,940B, BPFP=2.6201 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 27,072B, BPFP=1.5784 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 45,544B, BPFP=2.6553 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 39,036B, BPFP=2.2759 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 44,040B, BPFP=2.5676 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 95,580B, BPFP=0.7961 +⌛️ [2/4] FRONTEND: Frontend time: 0.354s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.output: torch.Size([1, 268, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.439s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 268, 128]) + layer.0.v_cache: torch.Size([1, 4, 268, 128]) + layer.1.k_cache: torch.Size([1, 4, 268, 128]) + layer.1.v_cache: torch.Size([1, 4, 268, 128]) + layer.2.k_cache: torch.Size([1, 4, 268, 128]) + layer.2.v_cache: torch.Size([1, 4, 268, 128]) + layer.3.k_cache: torch.Size([1, 4, 268, 128]) + layer.3.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.k_cache: torch.Size([1, 4, 268, 128]) + layer.4.v_cache: torch.Size([1, 4, 268, 128]) + layer.4.output: torch.Size([1, 268, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13150386 22.43552137 + layer.0.v_cache 0.00001401 0.00490178 + layer.1.k_cache 0.57442873 2.95112291 + layer.1.v_cache 0.00000554 0.00202834 + layer.2.k_cache 0.00779028 0.42189533 + layer.2.v_cache 0.00001875 0.00549934 + layer.3.k_cache 0.05176373 2.08171833 + layer.3.v_cache 0.00002027 0.00622672 + layer.4.k_cache 0.00069872 0.11541472 + layer.4.v_cache 0.00005023 0.01203049 + layer.4.output 0.01044900 202.82061234 + ------------------------------------------------------------------------------------- + TOTAL 0.04937866 85.16356739 + (elements=2,332,672) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 2332672 +Total Bytes 456156 +BPFP 1.5644 bits/point +EBPFP 3.1288 equivalent bits/point +MSE 85.163567 +---------------------- -------------------------------------------------------- +Time: 0.801s Load: 0.009s, Pack+Encode: 0.354s, Decode+Unpack: 0.439s +---------------------- -------------------------------------------------------- +💾 Converting with 85.1636 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/openbookqa/OpenBookQA-openbookqa-99.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/openbookqa/OpenBookQA-openbookqa-99.zst +------------------------ ---------------------------- +TOTAL PROCESSING SUMMARY +------------------------ ---------------------------- +Total files 100 +Avg BPFP 1.5347 bits/point +Avg EBPFP 3.0693 equivalent bits/point +Avg MSE 79.092358 +Avg Time 0.879s +------------------------ ---------------------------- diff --git a/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log b/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log new file mode 100644 index 0000000000000000000000000000000000000000..bc3fde65df849bff6c0815834d09a46e2d9a92ee --- /dev/null +++ b/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log @@ -0,0 +1,14258 @@ +Experiment: dtufc_hyperprior-featurecoding_kimiaudio_individual +Log file: output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/dtufc_hyperprior-featurecoding_kimiaudio_individual.log +DTUFCCodecConfig: + arch: hyperprior-featurecoding + handler: kimiaudio + checkpoint: codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar + transform_type: kmeans + transform_mapping:featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json + bit_depth: 8 + device: cuda:0 +Loading checkpoint: codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar +Checkpoint epoch: 598 +Loaded hyperprior-featurecoding (1-channel) on cuda:0 +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_feature.json: torch.Size([256]) +Loaded per-key quantization points for key 'output' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_feature.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.0.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.0.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_0_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.1.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.1.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_1_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.2.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.2.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_2_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.3.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.3.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_3_v.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_k.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.4.k_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_k.json +Loaded quantization points from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_v.json: torch.Size([256]) +Loaded per-key quantization points for key 'layer.4.v_cache' from featurecoding_utils/transform_mapping/kmeans10samples-8bits/kimiaudio/librispeech_fewshot-8bit_layer_4_v.json +Loaded per-key mappings: model=kimiaudio + Keys: ['output', 'layer.0.k_cache', 'layer.0.v_cache', 'layer.1.k_cache', 'layer.1.v_cache', 'layer.2.k_cache', 'layer.2.v_cache', 'layer.3.k_cache', 'layer.3.v_cache', 'layer.4.k_cache', 'layer.4.v_cache'] +---------------- ----------------------------------------------------------------------------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +Checkpoint codec_weights/hyperprior_hybrid/bmshj2018-hyperprior_lambda0.02_epochs600_lr0.0001_bs180_patch256-256_checkpoint_best.pth.tar +Transform type kmeans +Transform config featurecoding_utils/transform_mapping/kmeans10samples-8bits/mapping.json +Input ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa +Output output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa +---------------- ----------------------------------------------------------------------------------------------------------------------------- +Files found: 100 +---------------------------------------------------------------------- + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-1.zst (1/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-1.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 84, 128) +Output shape: (1, 84, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.output: torch.Size([1, 84, 3584]) -> torch.Size([1, 1, 84, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,128B, BPFP=0.9539 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,348B, BPFP=3.0409 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,884B, BPFP=1.6525 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,704B, BPFP=2.9211 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,112B, BPFP=1.8810 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,360B, BPFP=2.8571 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,716B, BPFP=1.8073 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,564B, BPFP=2.8951 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,576B, BPFP=2.5253 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,340B, BPFP=2.8534 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,328B, BPFP=0.9122 +⌛️ [2/4] FRONTEND: Frontend time: 0.647s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.311s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14073507 29.29952567 + layer.0.v_cache 0.00001426 0.00487686 + layer.1.k_cache 0.05042761 3.51975432 + layer.1.v_cache 0.00000517 0.00175123 + layer.2.k_cache 0.00244098 0.46472195 + layer.2.v_cache 0.00001715 0.00469888 + layer.3.k_cache 0.02726505 1.86985089 + layer.3.v_cache 0.00001796 0.00556283 + layer.4.k_cache 0.00069123 0.11546289 + layer.4.v_cache 0.00005084 0.01211583 + layer.4.output 0.16186065 627.51769770 + ------------------------------------------------------------------------------------- + TOTAL 0.07968764 260.46601207 + (elements=731,136) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 731136 +Total Bytes 160060 +BPFP 1.7514 bits/point +EBPFP 3.5027 equivalent bits/point +MSE 260.466012 +---------------------- -------------------------------------------------------- +Time: 0.962s Load: 0.005s, Pack+Encode: 0.647s, Decode+Unpack: 0.311s +---------------------- -------------------------------------------------------- +💾 Converting with 260.4660 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-1.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-1.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-10.zst (2/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-10.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 83, 128) +Output shape: (1, 83, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.0.v_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.1.k_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.1.v_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.2.k_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.2.v_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.3.k_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.3.v_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.4.k_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.4.v_cache: torch.Size([1, 4, 83, 128]) -> torch.Size([1, 1, 83, 512]) + layer.4.output: torch.Size([1, 83, 3584]) -> torch.Size([1, 1, 83, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,100B, BPFP=0.9601 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,828B, BPFP=2.9797 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,736B, BPFP=1.6446 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,116B, BPFP=2.8456 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,740B, BPFP=1.8336 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,676B, BPFP=2.7628 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,296B, BPFP=1.7500 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,976B, BPFP=2.8193 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,924B, BPFP=2.4330 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,580B, BPFP=2.7447 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,032B, BPFP=0.9152 +⌛️ [2/4] FRONTEND: Frontend time: 0.245s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 83, 128]) + layer.0.v_cache: torch.Size([1, 4, 83, 128]) + layer.1.k_cache: torch.Size([1, 4, 83, 128]) + layer.1.v_cache: torch.Size([1, 4, 83, 128]) + layer.2.k_cache: torch.Size([1, 4, 83, 128]) + layer.2.v_cache: torch.Size([1, 4, 83, 128]) + layer.3.k_cache: torch.Size([1, 4, 83, 128]) + layer.3.v_cache: torch.Size([1, 4, 83, 128]) + layer.4.k_cache: torch.Size([1, 4, 83, 128]) + layer.4.v_cache: torch.Size([1, 4, 83, 128]) + layer.4.output: torch.Size([1, 83, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.302s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 83, 128]) + layer.0.v_cache: torch.Size([1, 4, 83, 128]) + layer.1.k_cache: torch.Size([1, 4, 83, 128]) + layer.1.v_cache: torch.Size([1, 4, 83, 128]) + layer.2.k_cache: torch.Size([1, 4, 83, 128]) + layer.2.v_cache: torch.Size([1, 4, 83, 128]) + layer.3.k_cache: torch.Size([1, 4, 83, 128]) + layer.3.v_cache: torch.Size([1, 4, 83, 128]) + layer.4.k_cache: torch.Size([1, 4, 83, 128]) + layer.4.v_cache: torch.Size([1, 4, 83, 128]) + layer.4.output: torch.Size([1, 83, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11392649 27.48856951 + layer.0.v_cache 0.00001670 0.00447049 + layer.1.k_cache 0.05249626 3.58898632 + layer.1.v_cache 0.00000506 0.00163479 + layer.2.k_cache 0.00416583 0.43532810 + layer.2.v_cache 0.00001711 0.00463082 + layer.3.k_cache 0.05931453 2.12118567 + layer.3.v_cache 0.00001789 0.00531676 + layer.4.k_cache 0.00069248 0.11021164 + layer.4.v_cache 0.00004729 0.01097155 + layer.4.output 0.16383441 627.12435456 + ------------------------------------------------------------------------------------- + TOTAL 0.08103179 260.21422280 + (elements=722,432) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 722432 +Total Bytes 155004 +BPFP 1.7165 bits/point +EBPFP 3.4329 equivalent bits/point +MSE 260.214223 +---------------------- -------------------------------------------------------- +Time: 0.550s Load: 0.003s, Pack+Encode: 0.245s, Decode+Unpack: 0.302s +---------------------- -------------------------------------------------------- +💾 Converting with 260.2142 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-10.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-10.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-102.zst (3/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-102.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 94, 128) +Output shape: (1, 94, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.0.v_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.1.k_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.1.v_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.2.k_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.2.v_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.3.k_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.3.v_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.4.k_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.4.v_cache: torch.Size([1, 4, 94, 128]) -> torch.Size([1, 1, 94, 512]) + layer.4.output: torch.Size([1, 94, 3584]) -> torch.Size([1, 1, 94, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,532B, BPFP=0.9195 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,016B, BPFP=2.8285 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,644B, BPFP=1.6031 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,312B, BPFP=2.7114 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,696B, BPFP=1.7779 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,960B, BPFP=2.6529 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,024B, BPFP=1.6662 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,344B, BPFP=2.7168 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,252B, BPFP=2.3690 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,880B, BPFP=2.6396 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,520B, BPFP=0.7960 +⌛️ [2/4] FRONTEND: Frontend time: 0.250s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 94, 128]) + layer.0.v_cache: torch.Size([1, 4, 94, 128]) + layer.1.k_cache: torch.Size([1, 4, 94, 128]) + layer.1.v_cache: torch.Size([1, 4, 94, 128]) + layer.2.k_cache: torch.Size([1, 4, 94, 128]) + layer.2.v_cache: torch.Size([1, 4, 94, 128]) + layer.3.k_cache: torch.Size([1, 4, 94, 128]) + layer.3.v_cache: torch.Size([1, 4, 94, 128]) + layer.4.k_cache: torch.Size([1, 4, 94, 128]) + layer.4.v_cache: torch.Size([1, 4, 94, 128]) + layer.4.output: torch.Size([1, 94, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.282s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 94, 128]) + layer.0.v_cache: torch.Size([1, 4, 94, 128]) + layer.1.k_cache: torch.Size([1, 4, 94, 128]) + layer.1.v_cache: torch.Size([1, 4, 94, 128]) + layer.2.k_cache: torch.Size([1, 4, 94, 128]) + layer.2.v_cache: torch.Size([1, 4, 94, 128]) + layer.3.k_cache: torch.Size([1, 4, 94, 128]) + layer.3.v_cache: torch.Size([1, 4, 94, 128]) + layer.4.k_cache: torch.Size([1, 4, 94, 128]) + layer.4.v_cache: torch.Size([1, 4, 94, 128]) + layer.4.output: torch.Size([1, 94, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.17268721 31.08765168 + layer.0.v_cache 0.00001453 0.00483996 + layer.1.k_cache 0.08276152 3.26140383 + layer.1.v_cache 0.00000545 0.00184457 + layer.2.k_cache 0.00739249 0.48550399 + layer.2.v_cache 0.00001625 0.00472399 + layer.3.k_cache 0.05929063 2.18305417 + layer.3.v_cache 0.00001766 0.00563341 + layer.4.k_cache 0.00068371 0.11260548 + layer.4.v_cache 0.00005055 0.01142569 + layer.4.output 0.14466690 572.49468085 + ------------------------------------------------------------------------------------- + TOTAL 0.07856402 237.91890898 + (elements=818,176) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 818176 +Total Bytes 165180 +BPFP 1.6151 bits/point +EBPFP 3.2302 equivalent bits/point +MSE 237.918909 +---------------------- -------------------------------------------------------- +Time: 0.536s Load: 0.004s, Pack+Encode: 0.250s, Decode+Unpack: 0.282s +---------------------- -------------------------------------------------------- +💾 Converting with 237.9189 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-102.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-102.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-106.zst (4/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-106.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 112, 128) +Output shape: (1, 112, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.0.v_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.1.k_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.1.v_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.2.k_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.2.v_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.3.k_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.3.v_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.4.k_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.4.v_cache: torch.Size([1, 4, 112, 128]) -> torch.Size([1, 1, 112, 512]) + layer.4.output: torch.Size([1, 112, 3584]) -> torch.Size([1, 1, 112, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,996B, BPFP=0.8365 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,448B, BPFP=2.7132 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,956B, BPFP=1.5285 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,816B, BPFP=2.6250 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 12,312B, BPFP=1.7176 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,380B, BPFP=2.5642 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,484B, BPFP=1.6021 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,788B, BPFP=2.6211 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 16,136B, BPFP=2.2511 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 18,116B, BPFP=2.5273 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 44,472B, BPFP=0.8863 +⌛️ [2/4] FRONTEND: Frontend time: 0.222s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 112, 128]) + layer.0.v_cache: torch.Size([1, 4, 112, 128]) + layer.1.k_cache: torch.Size([1, 4, 112, 128]) + layer.1.v_cache: torch.Size([1, 4, 112, 128]) + layer.2.k_cache: torch.Size([1, 4, 112, 128]) + layer.2.v_cache: torch.Size([1, 4, 112, 128]) + layer.3.k_cache: torch.Size([1, 4, 112, 128]) + layer.3.v_cache: torch.Size([1, 4, 112, 128]) + layer.4.k_cache: torch.Size([1, 4, 112, 128]) + layer.4.v_cache: torch.Size([1, 4, 112, 128]) + layer.4.output: torch.Size([1, 112, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.295s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 112, 128]) + layer.0.v_cache: torch.Size([1, 4, 112, 128]) + layer.1.k_cache: torch.Size([1, 4, 112, 128]) + layer.1.v_cache: torch.Size([1, 4, 112, 128]) + layer.2.k_cache: torch.Size([1, 4, 112, 128]) + layer.2.v_cache: torch.Size([1, 4, 112, 128]) + layer.3.k_cache: torch.Size([1, 4, 112, 128]) + layer.3.v_cache: torch.Size([1, 4, 112, 128]) + layer.4.k_cache: torch.Size([1, 4, 112, 128]) + layer.4.v_cache: torch.Size([1, 4, 112, 128]) + layer.4.output: torch.Size([1, 112, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.17321212 27.47105190 + layer.0.v_cache 0.00001450 0.00475113 + layer.1.k_cache 0.08643453 3.38287571 + layer.1.v_cache 0.00000659 0.00188139 + layer.2.k_cache 0.00690480 0.40837829 + layer.2.v_cache 0.00001899 0.00478951 + layer.3.k_cache 0.01205242 2.15593106 + layer.3.v_cache 0.00001923 0.00569787 + layer.4.k_cache 0.00071999 0.10863738 + layer.4.v_cache 0.00004947 0.01064641 + layer.4.output 10.17892331 469.19989636 + ------------------------------------------------------------------------------------- + TOTAL 4.20775858 195.17375972 + (elements=974,848) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 974848 +Total Bytes 194904 +BPFP 1.5995 bits/point +EBPFP 3.1989 equivalent bits/point +MSE 195.173760 +---------------------- -------------------------------------------------------- +Time: 0.521s Load: 0.005s, Pack+Encode: 0.222s, Decode+Unpack: 0.295s +---------------------- -------------------------------------------------------- +💾 Converting with 195.1738 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-106.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-106.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-117.zst (5/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-117.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 88, 128) +Output shape: (1, 88, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.output: torch.Size([1, 88, 3584]) -> torch.Size([1, 1, 88, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,336B, BPFP=0.9474 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,420B, BPFP=2.9155 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,100B, BPFP=1.6158 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,824B, BPFP=2.8097 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,160B, BPFP=1.8040 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,292B, BPFP=2.7152 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,632B, BPFP=1.7102 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,620B, BPFP=2.7734 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,640B, BPFP=2.4219 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,156B, BPFP=2.6911 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,932B, BPFP=0.8353 +⌛️ [2/4] FRONTEND: Frontend time: 0.265s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.276s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11687154 29.45743075 + layer.0.v_cache 0.00001423 0.00494398 + layer.1.k_cache 0.05324238 3.37238381 + layer.1.v_cache 0.00000539 0.00185729 + layer.2.k_cache 0.00397935 0.46293233 + layer.2.v_cache 0.00001766 0.00466193 + layer.3.k_cache 0.01210797 2.19649904 + layer.3.v_cache 0.00001803 0.00591277 + layer.4.k_cache 0.00075037 0.11236237 + layer.4.v_cache 0.00004481 0.01094016 + layer.4.output 0.15448949 602.72854099 + ------------------------------------------------------------------------------------- + TOTAL 0.07461636 250.27821832 + (elements=765,952) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 765952 +Total Bytes 159112 +BPFP 1.6618 bits/point +EBPFP 3.3237 equivalent bits/point +MSE 250.278218 +---------------------- -------------------------------------------------------- +Time: 0.544s Load: 0.004s, Pack+Encode: 0.265s, Decode+Unpack: 0.276s +---------------------- -------------------------------------------------------- +💾 Converting with 250.2782 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-117.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-117.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-130.zst (6/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-130.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 81, 128) +Output shape: (1, 81, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.output: torch.Size([1, 81, 3584]) -> torch.Size([1, 1, 81, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,036B, BPFP=0.9715 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,936B, BPFP=3.0741 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,820B, BPFP=1.7014 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,284B, BPFP=2.9483 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,108B, BPFP=1.9498 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,900B, BPFP=2.8742 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,484B, BPFP=1.8295 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,140B, BPFP=2.9205 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,108B, BPFP=2.5285 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,788B, BPFP=2.8526 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,656B, BPFP=0.9550 +⌛️ [2/4] FRONTEND: Frontend time: 0.241s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.output: torch.Size([1, 81, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.259s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.output: torch.Size([1, 81, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14203883 29.05322868 + layer.0.v_cache 0.00001438 0.00546872 + layer.1.k_cache 0.05633916 3.87249341 + layer.1.v_cache 0.00000602 0.00227417 + layer.2.k_cache 0.00244657 0.49864329 + layer.2.v_cache 0.00001847 0.00557102 + layer.3.k_cache 0.05492775 2.53308916 + layer.3.v_cache 0.00001883 0.00633611 + layer.4.k_cache 0.00062984 0.11482865 + layer.4.v_cache 0.00005178 0.01313943 + layer.4.output 0.18538215 660.41396605 + ------------------------------------------------------------------------------------- + TOTAL 0.09142157 274.05899029 + (elements=705,024) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 705024 +Total Bytes 157260 +BPFP 1.7844 bits/point +EBPFP 3.5689 equivalent bits/point +MSE 274.058990 +---------------------- -------------------------------------------------------- +Time: 0.503s Load: 0.003s, Pack+Encode: 0.241s, Decode+Unpack: 0.259s +---------------------- -------------------------------------------------------- +💾 Converting with 274.0590 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-130.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-130.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-139.zst (7/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-139.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 90, 128) +Output shape: (1, 90, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.output: torch.Size([1, 90, 3584]) -> torch.Size([1, 1, 90, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,448B, BPFP=0.9458 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,632B, BPFP=2.8875 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,320B, BPFP=1.6181 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,876B, BPFP=2.7563 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,560B, BPFP=1.8333 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,656B, BPFP=2.7181 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,028B, BPFP=1.7410 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,060B, BPFP=2.7882 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,048B, BPFP=2.4389 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,464B, BPFP=2.6847 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 38,232B, BPFP=0.9482 +⌛️ [2/4] FRONTEND: Frontend time: 0.261s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.284s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13028670 28.87445204 + layer.0.v_cache 0.00001462 0.00458891 + layer.1.k_cache 0.04946813 3.34376560 + layer.1.v_cache 0.00000624 0.00167253 + layer.2.k_cache 0.00411291 0.44062856 + layer.2.v_cache 0.00001729 0.00446377 + layer.3.k_cache 0.01793825 2.20271691 + layer.3.v_cache 0.00001895 0.00575913 + layer.4.k_cache 0.00077120 0.11063290 + layer.4.v_cache 0.00004532 0.01036997 + layer.4.output 0.15109633 592.92718254 + ------------------------------------------------------------------------------------- + TOTAL 0.07413847 246.20525459 + (elements=783,360) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 783360 +Total Bytes 167324 +BPFP 1.7088 bits/point +EBPFP 3.4176 equivalent bits/point +MSE 246.205255 +---------------------- -------------------------------------------------------- +Time: 0.549s Load: 0.004s, Pack+Encode: 0.261s, Decode+Unpack: 0.284s +---------------------- -------------------------------------------------------- +💾 Converting with 246.2053 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-139.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-139.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-143.zst (8/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-143.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 57, 128) +Output shape: (1, 57, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.output: torch.Size([1, 57, 3584]) -> torch.Size([1, 1, 57, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,648B, BPFP=1.0000 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,496B, BPFP=2.8772 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,112B, BPFP=1.6754 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,188B, BPFP=2.7928 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,848B, BPFP=1.8772 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,768B, BPFP=2.6776 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,540B, BPFP=1.7928 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 10,100B, BPFP=2.7686 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,768B, BPFP=2.4035 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,828B, BPFP=2.6941 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,040B, BPFP=1.0197 +⌛️ [2/4] FRONTEND: Frontend time: 0.235s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.175s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11683119 32.05782920 + layer.0.v_cache 0.00001514 0.00641752 + layer.1.k_cache 0.01218004 3.65585434 + layer.1.v_cache 0.00000521 0.00218709 + layer.2.k_cache 0.00491684 0.51931020 + layer.2.v_cache 0.00001706 0.00595081 + layer.3.k_cache 0.10823302 2.46205005 + layer.3.v_cache 0.00002043 0.00751886 + layer.4.k_cache 0.00062874 0.12617659 + layer.4.v_cache 0.00004794 0.01343504 + layer.4.output 0.23833110 934.19086779 + ------------------------------------------------------------------------------------- + TOTAL 0.11242431 386.95251790 + (elements=496,128) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 496128 +Total Bytes 108336 +BPFP 1.7469 bits/point +EBPFP 3.4938 equivalent bits/point +MSE 386.952518 +---------------------- -------------------------------------------------------- +Time: 0.412s Load: 0.003s, Pack+Encode: 0.235s, Decode+Unpack: 0.175s +---------------------- -------------------------------------------------------- +💾 Converting with 386.9525 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-143.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-143.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-145.zst (9/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-145.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 50, 128) +Output shape: (1, 50, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.output: torch.Size([1, 50, 3584]) -> torch.Size([1, 1, 50, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,436B, BPFP=1.0737 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,724B, BPFP=3.0387 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,520B, BPFP=1.7250 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,252B, BPFP=2.8912 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,168B, BPFP=1.9275 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,084B, BPFP=2.8388 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,892B, BPFP=1.8413 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,388B, BPFP=2.9337 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,188B, BPFP=2.5587 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,268B, BPFP=2.8963 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 24,488B, BPFP=1.0932 +⌛️ [2/4] FRONTEND: Frontend time: 0.157s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.175s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14598725 35.00418945 + layer.0.v_cache 0.00001449 0.00621276 + layer.1.k_cache 0.01347294 3.67733185 + layer.1.v_cache 0.00000521 0.00215485 + layer.2.k_cache 0.00488209 0.59145432 + layer.2.v_cache 0.00001649 0.00574717 + layer.3.k_cache 0.04421538 2.19233704 + layer.3.v_cache 0.00001905 0.00745458 + layer.4.k_cache 0.00073779 0.12975303 + layer.4.v_cache 0.00005089 0.01234432 + layer.4.output 0.27158585 967.75464286 + ------------------------------------------------------------------------------------- + TOTAL 0.12414721 400.93596937 + (elements=435,200) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 435200 +Total Bytes 100408 +BPFP 1.8457 bits/point +EBPFP 3.6915 equivalent bits/point +MSE 400.935969 +---------------------- -------------------------------------------------------- +Time: 0.335s Load: 0.003s, Pack+Encode: 0.157s, Decode+Unpack: 0.175s +---------------------- -------------------------------------------------------- +💾 Converting with 400.9360 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-145.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-145.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-148.zst (10/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-148.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 46, 128) +Output shape: (1, 46, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.0.v_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.1.k_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.1.v_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.2.k_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.2.v_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.3.k_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.3.v_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.4.k_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.4.v_cache: torch.Size([1, 4, 46, 128]) -> torch.Size([1, 1, 46, 512]) + layer.4.output: torch.Size([1, 46, 3584]) -> torch.Size([1, 1, 46, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,256B, BPFP=1.1060 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,284B, BPFP=3.1535 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,272B, BPFP=1.7908 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,060B, BPFP=3.0774 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 5,864B, BPFP=1.9918 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 8,712B, BPFP=2.9592 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,612B, BPFP=1.9062 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 8,980B, BPFP=3.0503 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 7,860B, BPFP=2.6698 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 8,856B, BPFP=3.0082 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 23,200B, BPFP=1.1258 +⌛️ [2/4] FRONTEND: Frontend time: 0.143s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 46, 128]) + layer.0.v_cache: torch.Size([1, 4, 46, 128]) + layer.1.k_cache: torch.Size([1, 4, 46, 128]) + layer.1.v_cache: torch.Size([1, 4, 46, 128]) + layer.2.k_cache: torch.Size([1, 4, 46, 128]) + layer.2.v_cache: torch.Size([1, 4, 46, 128]) + layer.3.k_cache: torch.Size([1, 4, 46, 128]) + layer.3.v_cache: torch.Size([1, 4, 46, 128]) + layer.4.k_cache: torch.Size([1, 4, 46, 128]) + layer.4.v_cache: torch.Size([1, 4, 46, 128]) + layer.4.output: torch.Size([1, 46, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.166s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 46, 128]) + layer.0.v_cache: torch.Size([1, 4, 46, 128]) + layer.1.k_cache: torch.Size([1, 4, 46, 128]) + layer.1.v_cache: torch.Size([1, 4, 46, 128]) + layer.2.k_cache: torch.Size([1, 4, 46, 128]) + layer.2.v_cache: torch.Size([1, 4, 46, 128]) + layer.3.k_cache: torch.Size([1, 4, 46, 128]) + layer.3.v_cache: torch.Size([1, 4, 46, 128]) + layer.4.k_cache: torch.Size([1, 4, 46, 128]) + layer.4.v_cache: torch.Size([1, 4, 46, 128]) + layer.4.output: torch.Size([1, 46, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14129612 36.25816014 + layer.0.v_cache 0.00001532 0.00659273 + layer.1.k_cache 0.01315293 3.89909927 + layer.1.v_cache 0.00000553 0.00238096 + layer.2.k_cache 0.00779452 0.65259983 + layer.2.v_cache 0.00001723 0.00535431 + layer.3.k_cache 0.01828384 2.47931920 + layer.3.v_cache 0.00001944 0.00834422 + layer.4.k_cache 0.00060559 0.13265764 + layer.4.v_cache 0.00005113 0.01467922 + layer.4.output 0.29517000 1168.79211957 + ------------------------------------------------------------------------------------- + TOTAL 0.13220186 483.82376615 + (elements=400,384) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 400384 +Total Bytes 95956 +BPFP 1.9173 bits/point +EBPFP 3.8346 equivalent bits/point +MSE 483.823766 +---------------------- -------------------------------------------------------- +Time: 0.312s Load: 0.003s, Pack+Encode: 0.143s, Decode+Unpack: 0.166s +---------------------- -------------------------------------------------------- +💾 Converting with 483.8238 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-148.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-148.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-15.zst (11/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-15.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 110, 128) +Output shape: (1, 110, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.0.v_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.1.k_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.1.v_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.2.k_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.2.v_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.3.k_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.3.v_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.4.k_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.4.v_cache: torch.Size([1, 4, 110, 128]) -> torch.Size([1, 1, 110, 512]) + layer.4.output: torch.Size([1, 110, 3584]) -> torch.Size([1, 1, 110, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 6,040B, BPFP=0.8580 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,512B, BPFP=2.7716 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,932B, BPFP=1.5528 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,860B, BPFP=2.6790 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 12,184B, BPFP=1.7307 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,404B, BPFP=2.6142 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,636B, BPFP=1.6528 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,836B, BPFP=2.6756 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 16,144B, BPFP=2.2932 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 18,132B, BPFP=2.5756 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 42,760B, BPFP=0.8677 +⌛️ [2/4] FRONTEND: Frontend time: 0.277s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 110, 128]) + layer.0.v_cache: torch.Size([1, 4, 110, 128]) + layer.1.k_cache: torch.Size([1, 4, 110, 128]) + layer.1.v_cache: torch.Size([1, 4, 110, 128]) + layer.2.k_cache: torch.Size([1, 4, 110, 128]) + layer.2.v_cache: torch.Size([1, 4, 110, 128]) + layer.3.k_cache: torch.Size([1, 4, 110, 128]) + layer.3.v_cache: torch.Size([1, 4, 110, 128]) + layer.4.k_cache: torch.Size([1, 4, 110, 128]) + layer.4.v_cache: torch.Size([1, 4, 110, 128]) + layer.4.output: torch.Size([1, 110, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.297s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 110, 128]) + layer.0.v_cache: torch.Size([1, 4, 110, 128]) + layer.1.k_cache: torch.Size([1, 4, 110, 128]) + layer.1.v_cache: torch.Size([1, 4, 110, 128]) + layer.2.k_cache: torch.Size([1, 4, 110, 128]) + layer.2.v_cache: torch.Size([1, 4, 110, 128]) + layer.3.k_cache: torch.Size([1, 4, 110, 128]) + layer.3.v_cache: torch.Size([1, 4, 110, 128]) + layer.4.k_cache: torch.Size([1, 4, 110, 128]) + layer.4.v_cache: torch.Size([1, 4, 110, 128]) + layer.4.output: torch.Size([1, 110, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14476244 28.97654031 + layer.0.v_cache 0.00001476 0.00513402 + layer.1.k_cache 0.05665370 3.34524286 + layer.1.v_cache 0.00000525 0.00171386 + layer.2.k_cache 0.00786287 0.47687191 + layer.2.v_cache 0.00001801 0.00463767 + layer.3.k_cache 0.01410385 2.05906303 + layer.3.v_cache 0.00002027 0.00575461 + layer.4.k_cache 0.00066873 0.11064641 + layer.4.v_cache 0.00004626 0.01017429 + layer.4.output 10.36401748 484.53478084 + ------------------------------------------------------------------------------------- + TOTAL 4.28072226 201.57289676 + (elements=957,440) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 957440 +Total Bytes 193440 +BPFP 1.6163 bits/point +EBPFP 3.2326 equivalent bits/point +MSE 201.572897 +---------------------- -------------------------------------------------------- +Time: 0.578s Load: 0.005s, Pack+Encode: 0.277s, Decode+Unpack: 0.297s +---------------------- -------------------------------------------------------- +💾 Converting with 201.5729 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-15.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-15.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-156.zst (12/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-156.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 81, 128) +Output shape: (1, 81, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) -> torch.Size([1, 1, 81, 512]) + layer.4.output: torch.Size([1, 81, 3584]) -> torch.Size([1, 1, 81, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,060B, BPFP=0.9761 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,352B, BPFP=3.1543 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,896B, BPFP=1.7160 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,604B, BPFP=3.0100 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,372B, BPFP=2.0008 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,236B, BPFP=2.9390 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,864B, BPFP=1.9028 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,580B, BPFP=3.0054 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,596B, BPFP=2.6227 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,368B, BPFP=2.9645 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 42,020B, BPFP=1.1580 +⌛️ [2/4] FRONTEND: Frontend time: 0.236s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.output: torch.Size([1, 81, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.319s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 81, 128]) + layer.0.v_cache: torch.Size([1, 4, 81, 128]) + layer.1.k_cache: torch.Size([1, 4, 81, 128]) + layer.1.v_cache: torch.Size([1, 4, 81, 128]) + layer.2.k_cache: torch.Size([1, 4, 81, 128]) + layer.2.v_cache: torch.Size([1, 4, 81, 128]) + layer.3.k_cache: torch.Size([1, 4, 81, 128]) + layer.3.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.k_cache: torch.Size([1, 4, 81, 128]) + layer.4.v_cache: torch.Size([1, 4, 81, 128]) + layer.4.output: torch.Size([1, 81, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10533771 30.96794223 + layer.0.v_cache 0.00001610 0.00604667 + layer.1.k_cache 0.05553475 3.58785144 + layer.1.v_cache 0.00000748 0.00240922 + layer.2.k_cache 0.02166171 0.45324076 + layer.2.v_cache 0.00002000 0.00584476 + layer.3.k_cache 0.06386927 2.04224047 + layer.3.v_cache 0.00001944 0.00685024 + layer.4.k_cache 0.00064315 0.12575380 + layer.4.v_cache 0.00005385 0.01257532 + layer.4.output 0.16801396 653.02662037 + ------------------------------------------------------------------------------------- + TOTAL 0.08372125 271.08218221 + (elements=705,024) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 705024 +Total Bytes 167948 +BPFP 1.9057 bits/point +EBPFP 3.8115 equivalent bits/point +MSE 271.082182 +---------------------- -------------------------------------------------------- +Time: 0.559s Load: 0.004s, Pack+Encode: 0.236s, Decode+Unpack: 0.319s +---------------------- -------------------------------------------------------- +💾 Converting with 271.0822 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-156.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-156.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-158.zst (13/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-158.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 49, 128) +Output shape: (1, 49, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.0.v_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.1.k_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.1.v_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.2.k_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.2.v_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.3.k_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.3.v_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.4.k_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.4.v_cache: torch.Size([1, 4, 49, 128]) -> torch.Size([1, 1, 49, 512]) + layer.4.output: torch.Size([1, 49, 3584]) -> torch.Size([1, 1, 49, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,412B, BPFP=1.0880 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,568B, BPFP=3.0510 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,536B, BPFP=1.7653 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,316B, BPFP=2.9707 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,148B, BPFP=1.9605 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 8,992B, BPFP=2.8673 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,772B, BPFP=1.8406 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,280B, BPFP=2.9592 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,100B, BPFP=2.5829 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,144B, BPFP=2.9158 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,500B, BPFP=1.2072 +⌛️ [2/4] FRONTEND: Frontend time: 0.171s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 49, 128]) + layer.0.v_cache: torch.Size([1, 4, 49, 128]) + layer.1.k_cache: torch.Size([1, 4, 49, 128]) + layer.1.v_cache: torch.Size([1, 4, 49, 128]) + layer.2.k_cache: torch.Size([1, 4, 49, 128]) + layer.2.v_cache: torch.Size([1, 4, 49, 128]) + layer.3.k_cache: torch.Size([1, 4, 49, 128]) + layer.3.v_cache: torch.Size([1, 4, 49, 128]) + layer.4.k_cache: torch.Size([1, 4, 49, 128]) + layer.4.v_cache: torch.Size([1, 4, 49, 128]) + layer.4.output: torch.Size([1, 49, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.171s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 49, 128]) + layer.0.v_cache: torch.Size([1, 4, 49, 128]) + layer.1.k_cache: torch.Size([1, 4, 49, 128]) + layer.1.v_cache: torch.Size([1, 4, 49, 128]) + layer.2.k_cache: torch.Size([1, 4, 49, 128]) + layer.2.v_cache: torch.Size([1, 4, 49, 128]) + layer.3.k_cache: torch.Size([1, 4, 49, 128]) + layer.3.v_cache: torch.Size([1, 4, 49, 128]) + layer.4.k_cache: torch.Size([1, 4, 49, 128]) + layer.4.v_cache: torch.Size([1, 4, 49, 128]) + layer.4.output: torch.Size([1, 49, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12410851 35.59607432 + layer.0.v_cache 0.00001740 0.00667798 + layer.1.k_cache 0.01401279 3.81105446 + layer.1.v_cache 0.00000579 0.00251903 + layer.2.k_cache 0.00545398 0.51142942 + layer.2.v_cache 0.00001955 0.00682018 + layer.3.k_cache 0.07546502 2.43443236 + layer.3.v_cache 0.00001978 0.00788296 + layer.4.k_cache 0.00063206 0.13385509 + layer.4.v_cache 0.00005205 0.01333230 + layer.4.output 0.27729562 1044.34630102 + ------------------------------------------------------------------------------------- + TOTAL 0.12710919 432.52636384 + (elements=426,496) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 426496 +Total Bytes 101768 +BPFP 1.9089 bits/point +EBPFP 3.8178 equivalent bits/point +MSE 432.526364 +---------------------- -------------------------------------------------------- +Time: 0.345s Load: 0.003s, Pack+Encode: 0.171s, Decode+Unpack: 0.171s +---------------------- -------------------------------------------------------- +💾 Converting with 432.5264 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-158.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-158.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-161.zst (14/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-161.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 57, 128) +Output shape: (1, 57, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.output: torch.Size([1, 57, 3584]) -> torch.Size([1, 1, 57, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,624B, BPFP=0.9934 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,360B, BPFP=2.8399 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,020B, BPFP=1.6502 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,956B, BPFP=2.7292 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,868B, BPFP=1.8827 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,656B, BPFP=2.6469 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,348B, BPFP=1.7401 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,996B, BPFP=2.7401 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,724B, BPFP=2.3914 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,652B, BPFP=2.6458 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,648B, BPFP=1.0044 +⌛️ [2/4] FRONTEND: Frontend time: 0.151s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.165s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.19197876 33.24162212 + layer.0.v_cache 0.00001449 0.00641909 + layer.1.k_cache 0.01466701 3.78772829 + layer.1.v_cache 0.00000519 0.00219598 + layer.2.k_cache 0.00720649 0.53319653 + layer.2.v_cache 0.00001785 0.00550214 + layer.3.k_cache 0.10851213 2.42858084 + layer.3.v_cache 0.00001996 0.00724244 + layer.4.k_cache 0.00062085 0.12595225 + layer.4.v_cache 0.00004770 0.01266218 + layer.4.output 0.23832821 913.18953634 + ------------------------------------------------------------------------------------- + TOTAL 0.11714047 378.38105037 + (elements=496,128) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 496128 +Total Bytes 106852 +BPFP 1.7230 bits/point +EBPFP 3.4459 equivalent bits/point +MSE 378.381050 +---------------------- -------------------------------------------------------- +Time: 0.319s Load: 0.003s, Pack+Encode: 0.151s, Decode+Unpack: 0.165s +---------------------- -------------------------------------------------------- +💾 Converting with 378.3811 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-161.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-161.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-166.zst (15/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-166.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.002s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 51, 128) +Output shape: (1, 51, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.output: torch.Size([1, 51, 3584]) -> torch.Size([1, 1, 51, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,572B, BPFP=1.0944 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,944B, BPFP=3.0466 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,692B, BPFP=1.7439 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,516B, BPFP=2.9154 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,340B, BPFP=1.9424 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,244B, BPFP=2.8321 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,996B, BPFP=1.8370 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,532B, BPFP=2.9203 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,308B, BPFP=2.5453 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,440B, BPFP=2.8922 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,428B, BPFP=1.1129 +⌛️ [2/4] FRONTEND: Frontend time: 0.152s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.output: torch.Size([1, 51, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.161s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.output: torch.Size([1, 51, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12665685 31.89736998 + layer.0.v_cache 0.00001436 0.00580067 + layer.1.k_cache 0.01497617 3.75834386 + layer.1.v_cache 0.00000537 0.00220470 + layer.2.k_cache 0.00986420 0.45304889 + layer.2.v_cache 0.00001802 0.00569525 + layer.3.k_cache 0.09846177 2.53534055 + layer.3.v_cache 0.00001923 0.00717533 + layer.4.k_cache 0.00062397 0.12339521 + layer.4.v_cache 0.00005014 0.01292941 + layer.4.output 0.26631049 1020.75866597 + ------------------------------------------------------------------------------------- + TOTAL 0.12440374 422.59482151 + (elements=443,904) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 443904 +Total Bytes 103012 +BPFP 1.8565 bits/point +EBPFP 3.7129 equivalent bits/point +MSE 422.594822 +---------------------- -------------------------------------------------------- +Time: 0.316s Load: 0.002s, Pack+Encode: 0.152s, Decode+Unpack: 0.161s +---------------------- -------------------------------------------------------- +💾 Converting with 422.5948 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-166.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-166.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-182.zst (16/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-182.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 48, 128) +Output shape: (1, 48, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.output: torch.Size([1, 48, 3584]) -> torch.Size([1, 1, 48, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,344B, BPFP=1.0885 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,568B, BPFP=3.1146 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,476B, BPFP=1.7826 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,252B, BPFP=3.0117 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,188B, BPFP=2.0143 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,080B, BPFP=2.9557 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,960B, BPFP=1.9401 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,348B, BPFP=3.0430 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,288B, BPFP=2.6979 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,156B, BPFP=2.9805 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 27,704B, BPFP=1.2883 +⌛️ [2/4] FRONTEND: Frontend time: 0.148s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.output: torch.Size([1, 48, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.169s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.output: torch.Size([1, 48, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.18531116 35.88944753 + layer.0.v_cache 0.00001874 0.00692228 + layer.1.k_cache 0.01459585 4.19642417 + layer.1.v_cache 0.00000674 0.00290065 + layer.2.k_cache 0.01042949 0.53845914 + layer.2.v_cache 0.00001841 0.00699904 + layer.3.k_cache 0.15260293 2.30170170 + layer.3.v_cache 0.00002390 0.00920842 + layer.4.k_cache 0.00061422 0.13905440 + layer.4.v_cache 0.00004881 0.01477304 + layer.4.output 0.28297247 1100.64760045 + ------------------------------------------------------------------------------------- + TOTAL 0.13791044 455.74347609 + (elements=417,792) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 417792 +Total Bytes 103364 +BPFP 1.9792 bits/point +EBPFP 3.9585 equivalent bits/point +MSE 455.743476 +---------------------- -------------------------------------------------------- +Time: 0.320s Load: 0.003s, Pack+Encode: 0.148s, Decode+Unpack: 0.169s +---------------------- -------------------------------------------------------- +💾 Converting with 455.7435 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-182.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-182.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-192.zst (17/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-192.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 59, 128) +Output shape: (1, 59, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.output: torch.Size([1, 59, 3584]) -> torch.Size([1, 1, 59, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,816B, BPFP=1.0106 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,580B, BPFP=2.8019 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,260B, BPFP=1.6578 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,248B, BPFP=2.7140 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 7,088B, BPFP=1.8771 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,912B, BPFP=2.6250 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,768B, BPFP=1.7924 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 10,304B, BPFP=2.7288 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 9,032B, BPFP=2.3919 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 10,024B, BPFP=2.6547 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,244B, BPFP=1.1064 +⌛️ [2/4] FRONTEND: Frontend time: 0.153s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.output: torch.Size([1, 59, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.161s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.output: torch.Size([1, 59, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12052289 31.06827248 + layer.0.v_cache 0.00001546 0.00609657 + layer.1.k_cache 0.01434235 3.45902912 + layer.1.v_cache 0.00000574 0.00234967 + layer.2.k_cache 0.00254285 0.51652853 + layer.2.v_cache 0.00001710 0.00589552 + layer.3.k_cache 0.03737745 2.58379196 + layer.3.v_cache 0.00001946 0.00782595 + layer.4.k_cache 0.00063766 0.13068105 + layer.4.v_cache 0.00005039 0.01347194 + layer.4.output 0.23032015 897.11773608 + ------------------------------------------------------------------------------------- + TOTAL 0.10516308 371.62459384 + (elements=513,536) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 513536 +Total Bytes 113276 +BPFP 1.7646 bits/point +EBPFP 3.5293 equivalent bits/point +MSE 371.624594 +---------------------- -------------------------------------------------------- +Time: 0.318s Load: 0.003s, Pack+Encode: 0.153s, Decode+Unpack: 0.161s +---------------------- -------------------------------------------------------- +💾 Converting with 371.6246 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-192.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-192.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-207.zst (18/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-207.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 74, 128) +Output shape: (1, 74, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.output: torch.Size([1, 74, 3584]) -> torch.Size([1, 1, 74, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,976B, BPFP=1.0507 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,292B, BPFP=3.2289 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,456B, BPFP=1.7855 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,668B, BPFP=3.0971 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,936B, BPFP=2.0980 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,204B, BPFP=2.9992 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,096B, BPFP=1.9206 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,608B, BPFP=3.0845 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,640B, BPFP=2.6689 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,360B, BPFP=3.0321 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,500B, BPFP=1.0407 +⌛️ [2/4] FRONTEND: Frontend time: 0.241s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.276s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11335979 30.58272738 + layer.0.v_cache 0.00001513 0.00581137 + layer.1.k_cache 0.01655370 3.53120422 + layer.1.v_cache 0.00000667 0.00221693 + layer.2.k_cache 0.00428703 0.51766813 + layer.2.v_cache 0.00001789 0.00533441 + layer.3.k_cache 0.04938591 2.41373670 + layer.3.v_cache 0.00002111 0.00735472 + layer.4.k_cache 0.00066117 0.13316260 + layer.4.v_cache 0.00004716 0.01202220 + layer.4.output 0.18372514 718.20433156 + ------------------------------------------------------------------------------------- + TOTAL 0.08649597 297.92009174 + (elements=644,096) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 644096 +Total Bytes 152736 +BPFP 1.8971 bits/point +EBPFP 3.7941 equivalent bits/point +MSE 297.920092 +---------------------- -------------------------------------------------------- +Time: 0.521s Load: 0.004s, Pack+Encode: 0.241s, Decode+Unpack: 0.276s +---------------------- -------------------------------------------------------- +💾 Converting with 297.9201 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-207.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-207.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-208.zst (19/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-208.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 44, 128) +Output shape: (1, 44, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.output: torch.Size([1, 44, 3584]) -> torch.Size([1, 1, 44, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,248B, BPFP=1.1534 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,276B, BPFP=3.2940 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,316B, BPFP=1.8878 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 8,812B, BPFP=3.1293 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 5,976B, BPFP=2.1222 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 8,640B, BPFP=3.0682 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,712B, BPFP=2.0284 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 8,948B, BPFP=3.1776 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 7,868B, BPFP=2.7940 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 8,792B, BPFP=3.1222 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,752B, BPFP=1.3571 +⌛️ [2/4] FRONTEND: Frontend time: 0.151s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.output: torch.Size([1, 44, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.169s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.output: torch.Size([1, 44, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14777968 37.09550337 + layer.0.v_cache 0.00001531 0.00703532 + layer.1.k_cache 0.01728138 3.38432589 + layer.1.v_cache 0.00000547 0.00255375 + layer.2.k_cache 0.00823456 0.59035886 + layer.2.v_cache 0.00001932 0.00684575 + layer.3.k_cache 0.16770924 2.51709955 + layer.3.v_cache 0.00002046 0.00877437 + layer.4.k_cache 0.00061569 0.14265360 + layer.4.v_cache 0.00005592 0.01612001 + layer.4.output 0.30867824 1175.64194399 + ------------------------------------------------------------------------------------- + TOTAL 0.14720498 486.66263991 + (elements=382,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 382976 +Total Bytes 99340 +BPFP 2.0751 bits/point +EBPFP 4.1502 equivalent bits/point +MSE 486.662640 +---------------------- -------------------------------------------------------- +Time: 0.323s Load: 0.003s, Pack+Encode: 0.151s, Decode+Unpack: 0.169s +---------------------- -------------------------------------------------------- +💾 Converting with 486.6626 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-208.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-208.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-212.zst (20/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-212.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 78, 128) +Output shape: (1, 78, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.output: torch.Size([1, 78, 3584]) -> torch.Size([1, 1, 78, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,032B, BPFP=1.0080 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,248B, BPFP=3.2548 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,816B, BPFP=1.7660 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,660B, BPFP=3.1370 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,428B, BPFP=2.0889 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,160B, BPFP=3.0369 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,896B, BPFP=1.9824 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,652B, BPFP=3.1354 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,368B, BPFP=2.6779 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,180B, BPFP=3.0409 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 35,924B, BPFP=1.0280 +⌛️ [2/4] FRONTEND: Frontend time: 0.247s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.output: torch.Size([1, 78, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.283s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.output: torch.Size([1, 78, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14408422 31.20000438 + layer.0.v_cache 0.00002097 0.00662049 + layer.1.k_cache 0.05735342 3.45312304 + layer.1.v_cache 0.00000575 0.00246617 + layer.2.k_cache 0.01678689 0.54979500 + layer.2.v_cache 0.00001894 0.00628058 + layer.3.k_cache 0.03029843 2.31103672 + layer.3.v_cache 0.00002125 0.00803070 + layer.4.k_cache 0.00062582 0.13102530 + layer.4.v_cache 0.00005050 0.01486946 + layer.4.output 0.17433185 688.52896062 + ------------------------------------------------------------------------------------- + TOTAL 0.08644642 285.72858684 + (elements=678,912) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 678912 +Total Bytes 161364 +BPFP 1.9014 bits/point +EBPFP 3.8029 equivalent bits/point +MSE 285.728587 +---------------------- -------------------------------------------------------- +Time: 0.533s Load: 0.003s, Pack+Encode: 0.247s, Decode+Unpack: 0.283s +---------------------- -------------------------------------------------------- +💾 Converting with 285.7286 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-212.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-212.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-215.zst (21/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-215.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.002s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 44, 128) +Output shape: (1, 44, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) -> torch.Size([1, 1, 44, 512]) + layer.4.output: torch.Size([1, 44, 3584]) -> torch.Size([1, 1, 44, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,248B, BPFP=1.1534 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,060B, BPFP=3.2173 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,268B, BPFP=1.8707 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 8,700B, BPFP=3.0895 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 5,812B, BPFP=2.0639 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 8,552B, BPFP=3.0369 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,516B, BPFP=1.9588 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 8,760B, BPFP=3.1108 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 7,784B, BPFP=2.7642 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 8,684B, BPFP=3.0838 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 23,484B, BPFP=1.1914 +⌛️ [2/4] FRONTEND: Frontend time: 0.155s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.output: torch.Size([1, 44, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.167s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 44, 128]) + layer.0.v_cache: torch.Size([1, 4, 44, 128]) + layer.1.k_cache: torch.Size([1, 4, 44, 128]) + layer.1.v_cache: torch.Size([1, 4, 44, 128]) + layer.2.k_cache: torch.Size([1, 4, 44, 128]) + layer.2.v_cache: torch.Size([1, 4, 44, 128]) + layer.3.k_cache: torch.Size([1, 4, 44, 128]) + layer.3.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.k_cache: torch.Size([1, 4, 44, 128]) + layer.4.v_cache: torch.Size([1, 4, 44, 128]) + layer.4.output: torch.Size([1, 44, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13243668 36.64967485 + layer.0.v_cache 0.00001576 0.00660939 + layer.1.k_cache 0.01981235 3.75451279 + layer.1.v_cache 0.00000577 0.00256927 + layer.2.k_cache 0.01756107 0.60313641 + layer.2.v_cache 0.00001728 0.00612316 + layer.3.k_cache 0.04920139 2.56782931 + layer.3.v_cache 0.00001837 0.00754753 + layer.4.k_cache 0.00061593 0.14299108 + layer.4.v_cache 0.00004907 0.01469519 + layer.4.output 0.30854983 1177.17116477 + ------------------------------------------------------------------------------------- + TOTAL 0.13997544 487.29140249 + (elements=382,976) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 382976 +Total Bytes 94868 +BPFP 1.9817 bits/point +EBPFP 3.9634 equivalent bits/point +MSE 487.291402 +---------------------- -------------------------------------------------------- +Time: 0.324s Load: 0.002s, Pack+Encode: 0.155s, Decode+Unpack: 0.167s +---------------------- -------------------------------------------------------- +💾 Converting with 487.2914 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-215.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-215.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-217.zst (22/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-217.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 50, 128) +Output shape: (1, 50, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.output: torch.Size([1, 50, 3584]) -> torch.Size([1, 1, 50, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,424B, BPFP=1.0700 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,860B, BPFP=3.0812 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,632B, BPFP=1.7600 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,612B, BPFP=3.0038 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,356B, BPFP=1.9863 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,312B, BPFP=2.9100 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,016B, BPFP=1.8800 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,584B, BPFP=2.9950 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,268B, BPFP=2.5838 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,428B, BPFP=2.9463 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,476B, BPFP=1.1373 +⌛️ [2/4] FRONTEND: Frontend time: 0.164s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.169s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.20666794 35.99182861 + layer.0.v_cache 0.00002565 0.00654832 + layer.1.k_cache 0.01518770 3.61840759 + layer.1.v_cache 0.00000581 0.00242239 + layer.2.k_cache 0.01031687 0.51893410 + layer.2.v_cache 0.00001873 0.00653511 + layer.3.k_cache 0.09638391 2.38637634 + layer.3.v_cache 0.00001967 0.00822367 + layer.4.k_cache 0.00062172 0.13662227 + layer.4.v_cache 0.00004827 0.01430179 + layer.4.output 0.27162739 1065.47767857 + ------------------------------------------------------------------------------------- + TOTAL 0.13121694 441.23729119 + (elements=435,200) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 435200 +Total Bytes 102968 +BPFP 1.8928 bits/point +EBPFP 3.7856 equivalent bits/point +MSE 441.237291 +---------------------- -------------------------------------------------------- +Time: 0.335s Load: 0.003s, Pack+Encode: 0.164s, Decode+Unpack: 0.169s +---------------------- -------------------------------------------------------- +💾 Converting with 441.2373 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-217.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-217.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-219.zst (23/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-219.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 68, 128) +Output shape: (1, 68, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.output: torch.Size([1, 68, 3584]) -> torch.Size([1, 1, 68, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,592B, BPFP=1.0551 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,680B, BPFP=3.3732 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,204B, BPFP=1.8851 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,744B, BPFP=3.1581 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,276B, BPFP=2.1314 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,104B, BPFP=3.0110 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,800B, BPFP=2.0221 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,544B, BPFP=3.1121 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,808B, BPFP=2.7132 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,652B, BPFP=3.1369 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 36,092B, BPFP=1.1847 +⌛️ [2/4] FRONTEND: Frontend time: 0.249s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.292s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13431002 29.73650764 + layer.0.v_cache 0.00001720 0.00642863 + layer.1.k_cache 0.01179634 3.64144202 + layer.1.v_cache 0.00000537 0.00220868 + layer.2.k_cache 0.00798416 0.53348597 + layer.2.v_cache 0.00001783 0.00595119 + layer.3.k_cache 0.05795463 2.32881165 + layer.3.v_cache 0.00001974 0.00752642 + layer.4.k_cache 0.00062059 0.12531460 + layer.4.v_cache 0.00005204 0.01366301 + layer.4.output 0.19989206 761.72997637 + ------------------------------------------------------------------------------------- + TOTAL 0.09482484 315.79477496 + (elements=591,872) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 591872 +Total Bytes 147496 +BPFP 1.9936 bits/point +EBPFP 3.9872 equivalent bits/point +MSE 315.794775 +---------------------- -------------------------------------------------------- +Time: 0.545s Load: 0.004s, Pack+Encode: 0.249s, Decode+Unpack: 0.292s +---------------------- -------------------------------------------------------- +💾 Converting with 315.7948 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-219.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-219.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-224.zst (24/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-224.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.002s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 50, 128) +Output shape: (1, 50, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) -> torch.Size([1, 1, 50, 512]) + layer.4.output: torch.Size([1, 50, 3584]) -> torch.Size([1, 1, 50, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,440B, BPFP=1.0750 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,892B, BPFP=3.0913 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,600B, BPFP=1.7500 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,480B, BPFP=2.9625 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,332B, BPFP=1.9788 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,300B, BPFP=2.9062 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,020B, BPFP=1.8813 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,588B, BPFP=2.9962 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,324B, BPFP=2.6012 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,480B, BPFP=2.9625 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,384B, BPFP=1.1332 +⌛️ [2/4] FRONTEND: Frontend time: 0.159s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.188s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 50, 128]) + layer.0.v_cache: torch.Size([1, 4, 50, 128]) + layer.1.k_cache: torch.Size([1, 4, 50, 128]) + layer.1.v_cache: torch.Size([1, 4, 50, 128]) + layer.2.k_cache: torch.Size([1, 4, 50, 128]) + layer.2.v_cache: torch.Size([1, 4, 50, 128]) + layer.3.k_cache: torch.Size([1, 4, 50, 128]) + layer.3.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.k_cache: torch.Size([1, 4, 50, 128]) + layer.4.v_cache: torch.Size([1, 4, 50, 128]) + layer.4.output: torch.Size([1, 50, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15467889 34.65533936 + layer.0.v_cache 0.00001741 0.00643477 + layer.1.k_cache 0.01620592 3.81971039 + layer.1.v_cache 0.00000548 0.00227935 + layer.2.k_cache 0.00723796 0.49158352 + layer.2.v_cache 0.00001737 0.00613874 + layer.3.k_cache 0.06661982 2.55740952 + layer.3.v_cache 0.00002040 0.00769085 + layer.4.k_cache 0.00058776 0.12944824 + layer.4.v_cache 0.00004656 0.01333608 + layer.4.output 0.27162074 1064.73348214 + ------------------------------------------------------------------------------------- + TOTAL 0.12628134 440.87198505 + (elements=435,200) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 435200 +Total Bytes 102840 +BPFP 1.8904 bits/point +EBPFP 3.7809 equivalent bits/point +MSE 440.871985 +---------------------- -------------------------------------------------------- +Time: 0.349s Load: 0.002s, Pack+Encode: 0.159s, Decode+Unpack: 0.188s +---------------------- -------------------------------------------------------- +💾 Converting with 440.8720 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-224.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-224.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-225.zst (25/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-225.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 51, 128) +Output shape: (1, 51, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) -> torch.Size([1, 1, 51, 512]) + layer.4.output: torch.Size([1, 51, 3584]) -> torch.Size([1, 1, 51, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,568B, BPFP=1.0931 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,956B, BPFP=3.0502 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,660B, BPFP=1.7341 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,516B, BPFP=2.9154 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,300B, BPFP=1.9301 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,276B, BPFP=2.8419 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,984B, BPFP=1.8333 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,652B, BPFP=2.9571 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,308B, BPFP=2.5453 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,392B, BPFP=2.8775 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 24,452B, BPFP=1.0702 +⌛️ [2/4] FRONTEND: Frontend time: 0.165s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.output: torch.Size([1, 51, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.182s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 51, 128]) + layer.0.v_cache: torch.Size([1, 4, 51, 128]) + layer.1.k_cache: torch.Size([1, 4, 51, 128]) + layer.1.v_cache: torch.Size([1, 4, 51, 128]) + layer.2.k_cache: torch.Size([1, 4, 51, 128]) + layer.2.v_cache: torch.Size([1, 4, 51, 128]) + layer.3.k_cache: torch.Size([1, 4, 51, 128]) + layer.3.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.k_cache: torch.Size([1, 4, 51, 128]) + layer.4.v_cache: torch.Size([1, 4, 51, 128]) + layer.4.output: torch.Size([1, 51, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11261671 32.83688055 + layer.0.v_cache 0.00001829 0.00582533 + layer.1.k_cache 0.01190703 3.32751944 + layer.1.v_cache 0.00000528 0.00210212 + layer.2.k_cache 0.00973116 0.47582746 + layer.2.v_cache 0.00001668 0.00561115 + layer.3.k_cache 0.04200072 2.09479957 + layer.3.v_cache 0.00001874 0.00771247 + layer.4.k_cache 0.00062040 0.12493758 + layer.4.v_cache 0.00004776 0.01329034 + layer.4.output 0.26626884 1022.74150910 + ------------------------------------------------------------------------------------- + TOTAL 0.12005086 423.41676881 + (elements=443,904) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 443904 +Total Bytes 102064 +BPFP 1.8394 bits/point +EBPFP 3.6788 equivalent bits/point +MSE 423.416769 +---------------------- -------------------------------------------------------- +Time: 0.351s Load: 0.003s, Pack+Encode: 0.165s, Decode+Unpack: 0.182s +---------------------- -------------------------------------------------------- +💾 Converting with 423.4168 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-225.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-225.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-229.zst (26/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-229.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 52, 128) +Output shape: (1, 52, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.output: torch.Size([1, 52, 3584]) -> torch.Size([1, 1, 52, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,632B, BPFP=1.0913 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,232B, BPFP=3.0745 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,848B, BPFP=1.7572 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,836B, BPFP=2.9555 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,624B, BPFP=1.9904 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,584B, BPFP=2.8798 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,252B, BPFP=1.8786 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,912B, BPFP=2.9784 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,648B, BPFP=2.5986 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,716B, BPFP=2.9195 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,748B, BPFP=1.2340 +⌛️ [2/4] FRONTEND: Frontend time: 0.151s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.162s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12932784 32.44670692 + layer.0.v_cache 0.00001491 0.00692693 + layer.1.k_cache 0.01488123 3.67645792 + layer.1.v_cache 0.00000563 0.00245569 + layer.2.k_cache 0.00939683 0.50916217 + layer.2.v_cache 0.00001981 0.00667340 + layer.3.k_cache 0.07051506 2.47810657 + layer.3.v_cache 0.00001927 0.00802896 + layer.4.k_cache 0.00064413 0.13681295 + layer.4.v_cache 0.00005939 0.01480653 + layer.4.output 0.26127321 997.95501374 + ------------------------------------------------------------------------------------- + TOTAL 0.12081156 413.23360201 + (elements=452,608) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 452608 +Total Bytes 109032 +BPFP 1.9272 bits/point +EBPFP 3.8544 equivalent bits/point +MSE 413.233602 +---------------------- -------------------------------------------------------- +Time: 0.315s Load: 0.003s, Pack+Encode: 0.151s, Decode+Unpack: 0.162s +---------------------- -------------------------------------------------------- +💾 Converting with 413.2336 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-229.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-229.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-235.zst (27/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-235.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.002s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 52, 128) +Output shape: (1, 52, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.output: torch.Size([1, 52, 3584]) -> torch.Size([1, 1, 52, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,684B, BPFP=1.1070 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,308B, BPFP=3.0974 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,888B, BPFP=1.7692 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,012B, BPFP=3.0084 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,672B, BPFP=2.0048 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,792B, BPFP=2.9423 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,224B, BPFP=1.8702 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,860B, BPFP=2.9627 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,608B, BPFP=2.5865 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,740B, BPFP=2.9267 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 27,832B, BPFP=1.1947 +⌛️ [2/4] FRONTEND: Frontend time: 0.160s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.168s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13190371 34.49402090 + layer.0.v_cache 0.00001552 0.00676073 + layer.1.k_cache 0.01668521 4.05690384 + layer.1.v_cache 0.00000581 0.00263480 + layer.2.k_cache 0.00490881 0.49790338 + layer.2.v_cache 0.00001873 0.00646790 + layer.3.k_cache 0.11824917 2.39782055 + layer.3.v_cache 0.00002115 0.00787651 + layer.4.k_cache 0.00061206 0.13742410 + layer.4.v_cache 0.00004844 0.01355488 + layer.4.output 0.26122982 995.47278503 + ------------------------------------------------------------------------------------- + TOTAL 0.12359279 412.34887428 + (elements=452,608) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 452608 +Total Bytes 108620 +BPFP 1.9199 bits/point +EBPFP 3.8398 equivalent bits/point +MSE 412.348874 +---------------------- -------------------------------------------------------- +Time: 0.330s Load: 0.002s, Pack+Encode: 0.160s, Decode+Unpack: 0.168s +---------------------- -------------------------------------------------------- +💾 Converting with 412.3489 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-235.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-235.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-238.zst (28/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-238.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 92, 128) +Output shape: (1, 92, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.output: torch.Size([1, 92, 3584]) -> torch.Size([1, 1, 92, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,612B, BPFP=0.9531 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,064B, BPFP=2.8981 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,708B, BPFP=1.6488 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,488B, BPFP=2.8003 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,836B, BPFP=1.8404 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,140B, BPFP=2.7412 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,324B, BPFP=1.7534 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,564B, BPFP=2.8132 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,312B, BPFP=2.4307 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,880B, BPFP=2.6970 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 41,040B, BPFP=0.9957 +⌛️ [2/4] FRONTEND: Frontend time: 0.211s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.283s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10326390 29.40384575 + layer.0.v_cache 0.00001541 0.00522377 + layer.1.k_cache 0.04976616 3.35160794 + layer.1.v_cache 0.00000534 0.00180422 + layer.2.k_cache 0.01168035 0.49281369 + layer.2.v_cache 0.00001798 0.00485363 + layer.3.k_cache 0.02907468 1.98388639 + layer.3.v_cache 0.00001946 0.00613903 + layer.4.k_cache 0.00073861 0.11706689 + layer.4.v_cache 0.00005077 0.01156296 + layer.4.output 0.14788401 572.28838315 + ------------------------------------------------------------------------------------- + TOTAL 0.07234240 237.72926390 + (elements=800,768) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 800768 +Total Bytes 173968 +BPFP 1.7380 bits/point +EBPFP 3.4760 equivalent bits/point +MSE 237.729264 +---------------------- -------------------------------------------------------- +Time: 0.499s Load: 0.005s, Pack+Encode: 0.211s, Decode+Unpack: 0.283s +---------------------- -------------------------------------------------------- +💾 Converting with 237.7293 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-238.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-238.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-24.zst (29/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-24.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 64, 128) +Output shape: (1, 64, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.0.v_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.1.k_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.1.v_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.2.k_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.2.v_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.3.k_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.3.v_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.4.k_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.4.v_cache: torch.Size([1, 4, 64, 128]) -> torch.Size([1, 1, 64, 512]) + layer.4.output: torch.Size([1, 64, 3584]) -> torch.Size([1, 1, 64, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,224B, BPFP=0.7871 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,128B, BPFP=2.4727 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,560B, BPFP=1.3574 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,788B, BPFP=2.3896 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,304B, BPFP=1.5391 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,636B, BPFP=2.3525 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,972B, BPFP=1.4580 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,832B, BPFP=2.4004 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,452B, BPFP=2.0635 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,484B, BPFP=2.3154 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,716B, BPFP=1.0015 +⌛️ [2/4] FRONTEND: Frontend time: 0.149s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 64, 128]) + layer.0.v_cache: torch.Size([1, 4, 64, 128]) + layer.1.k_cache: torch.Size([1, 4, 64, 128]) + layer.1.v_cache: torch.Size([1, 4, 64, 128]) + layer.2.k_cache: torch.Size([1, 4, 64, 128]) + layer.2.v_cache: torch.Size([1, 4, 64, 128]) + layer.3.k_cache: torch.Size([1, 4, 64, 128]) + layer.3.v_cache: torch.Size([1, 4, 64, 128]) + layer.4.k_cache: torch.Size([1, 4, 64, 128]) + layer.4.v_cache: torch.Size([1, 4, 64, 128]) + layer.4.output: torch.Size([1, 64, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.161s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 64, 128]) + layer.0.v_cache: torch.Size([1, 4, 64, 128]) + layer.1.k_cache: torch.Size([1, 4, 64, 128]) + layer.1.v_cache: torch.Size([1, 4, 64, 128]) + layer.2.k_cache: torch.Size([1, 4, 64, 128]) + layer.2.v_cache: torch.Size([1, 4, 64, 128]) + layer.3.k_cache: torch.Size([1, 4, 64, 128]) + layer.3.v_cache: torch.Size([1, 4, 64, 128]) + layer.4.k_cache: torch.Size([1, 4, 64, 128]) + layer.4.v_cache: torch.Size([1, 4, 64, 128]) + layer.4.output: torch.Size([1, 64, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09722494 22.89024353 + layer.0.v_cache 0.00001391 0.00516774 + layer.1.k_cache 0.01806639 2.94336200 + layer.1.v_cache 0.00000557 0.00201420 + layer.2.k_cache 0.00423524 0.34775695 + layer.2.v_cache 0.00001868 0.00554561 + layer.3.k_cache 0.03759564 2.02632737 + layer.3.v_cache 0.00001825 0.00615092 + layer.4.k_cache 0.00061753 0.10703606 + layer.4.v_cache 0.00004894 0.01219026 + layer.4.output 0.21408452 723.68415179 + ------------------------------------------------------------------------------------- + TOTAL 0.09743746 299.65499160 + (elements=557,056) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 557056 +Total Bytes 107096 +BPFP 1.5380 bits/point +EBPFP 3.0761 equivalent bits/point +MSE 299.654992 +---------------------- -------------------------------------------------------- +Time: 0.313s Load: 0.003s, Pack+Encode: 0.149s, Decode+Unpack: 0.161s +---------------------- -------------------------------------------------------- +💾 Converting with 299.6550 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-24.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-24.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-263.zst (30/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-263.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 102, 128) +Output shape: (1, 102, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.output: torch.Size([1, 102, 3584]) -> torch.Size([1, 1, 102, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,772B, BPFP=0.8842 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 18,516B, BPFP=2.8364 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,248B, BPFP=1.5699 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 17,856B, BPFP=2.7353 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,404B, BPFP=1.7469 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 17,308B, BPFP=2.6513 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,732B, BPFP=1.6440 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 17,660B, BPFP=2.7053 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 15,428B, BPFP=2.3634 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 17,152B, BPFP=2.6275 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 38,132B, BPFP=0.8345 +⌛️ [2/4] FRONTEND: Frontend time: 0.266s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.output: torch.Size([1, 102, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.288s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.output: torch.Size([1, 102, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11648470 26.70461857 + layer.0.v_cache 0.00001454 0.00471563 + layer.1.k_cache 0.02721847 3.57580207 + layer.1.v_cache 0.00000531 0.00177353 + layer.2.k_cache 0.01045542 0.44996247 + layer.2.v_cache 0.00001699 0.00450476 + layer.3.k_cache 0.01744144 1.86973662 + layer.3.v_cache 0.00001839 0.00555049 + layer.4.k_cache 0.00076342 0.11268947 + layer.4.v_cache 0.00004795 0.01073585 + layer.4.output 11.17676328 522.17765231 + ------------------------------------------------------------------------------------- + TOTAL 4.61234174 216.94021504 + (elements=887,808) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 887808 +Total Bytes 180208 +BPFP 1.6238 bits/point +EBPFP 3.2477 equivalent bits/point +MSE 216.940215 +---------------------- -------------------------------------------------------- +Time: 0.558s Load: 0.004s, Pack+Encode: 0.266s, Decode+Unpack: 0.288s +---------------------- -------------------------------------------------------- +💾 Converting with 216.9402 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-263.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-263.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-264.zst (31/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-264.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 80, 128) +Output shape: (1, 80, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.output: torch.Size([1, 80, 3584]) -> torch.Size([1, 1, 80, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,948B, BPFP=0.9664 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,636B, BPFP=3.0539 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,464B, BPFP=1.6531 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,864B, BPFP=2.9031 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,572B, BPFP=1.8695 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,456B, BPFP=2.8234 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,164B, BPFP=1.7898 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,824B, BPFP=2.8953 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,820B, BPFP=2.5039 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,424B, BPFP=2.8172 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,332B, BPFP=0.9021 +⌛️ [2/4] FRONTEND: Frontend time: 0.230s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.304s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11433966 29.60415649 + layer.0.v_cache 0.00001437 0.00493812 + layer.1.k_cache 0.05315235 3.73015785 + layer.1.v_cache 0.00000524 0.00164882 + layer.2.k_cache 0.00678293 0.45723004 + layer.2.v_cache 0.00001704 0.00457195 + layer.3.k_cache 0.01951189 2.36069775 + layer.3.v_cache 0.00001758 0.00554728 + layer.4.k_cache 0.00076624 0.11328874 + layer.4.v_cache 0.00004856 0.01096895 + layer.4.output 0.16992235 663.44916295 + ------------------------------------------------------------------------------------- + TOTAL 0.08141837 275.31984392 + (elements=696,320) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 696320 +Total Bytes 151504 +BPFP 1.7406 bits/point +EBPFP 3.4813 equivalent bits/point +MSE 275.319844 +---------------------- -------------------------------------------------------- +Time: 0.540s Load: 0.005s, Pack+Encode: 0.230s, Decode+Unpack: 0.304s +---------------------- -------------------------------------------------------- +💾 Converting with 275.3198 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-264.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-264.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-271.zst (32/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-271.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 90, 128) +Output shape: (1, 90, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.output: torch.Size([1, 90, 3584]) -> torch.Size([1, 1, 90, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,420B, BPFP=0.9410 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,764B, BPFP=2.9104 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,320B, BPFP=1.6181 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,092B, BPFP=2.7938 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,696B, BPFP=1.8569 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,744B, BPFP=2.7333 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,168B, BPFP=1.7653 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,256B, BPFP=2.8222 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,248B, BPFP=2.4736 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,716B, BPFP=2.7285 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 39,976B, BPFP=0.9915 +⌛️ [2/4] FRONTEND: Frontend time: 0.227s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.269s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11570581 28.19070638 + layer.0.v_cache 0.00001517 0.00490871 + layer.1.k_cache 0.03236232 3.45370721 + layer.1.v_cache 0.00000662 0.00180924 + layer.2.k_cache 0.00641356 0.44166251 + layer.2.v_cache 0.00001676 0.00469613 + layer.3.k_cache 0.02648412 2.29375983 + layer.3.v_cache 0.00001896 0.00619740 + layer.4.k_cache 0.00074524 0.11707464 + layer.4.v_cache 0.00004843 0.01149741 + layer.4.output 0.15113260 593.06180556 + ------------------------------------------------------------------------------------- + TOTAL 0.07292619 246.23286226 + (elements=783,360) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 783360 +Total Bytes 170400 +BPFP 1.7402 bits/point +EBPFP 3.4804 equivalent bits/point +MSE 246.232862 +---------------------- -------------------------------------------------------- +Time: 0.501s Load: 0.004s, Pack+Encode: 0.227s, Decode+Unpack: 0.269s +---------------------- -------------------------------------------------------- +💾 Converting with 246.2329 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-271.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-271.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-275.zst (33/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-275.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 114, 128) +Output shape: (1, 114, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.0.v_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.1.k_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.1.v_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.2.k_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.2.v_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.3.k_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.3.v_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.4.k_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.4.v_cache: torch.Size([1, 4, 114, 128]) -> torch.Size([1, 1, 114, 512]) + layer.4.output: torch.Size([1, 114, 3584]) -> torch.Size([1, 1, 114, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 6,072B, BPFP=0.8322 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,448B, BPFP=2.6656 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,672B, BPFP=1.4627 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,680B, BPFP=2.5603 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 12,004B, BPFP=1.6453 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,252B, BPFP=2.5016 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,372B, BPFP=1.5587 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,652B, BPFP=2.5565 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 16,200B, BPFP=2.2204 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 18,148B, BPFP=2.4874 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 42,516B, BPFP=0.8325 +⌛️ [2/4] FRONTEND: Frontend time: 0.254s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 114, 128]) + layer.0.v_cache: torch.Size([1, 4, 114, 128]) + layer.1.k_cache: torch.Size([1, 4, 114, 128]) + layer.1.v_cache: torch.Size([1, 4, 114, 128]) + layer.2.k_cache: torch.Size([1, 4, 114, 128]) + layer.2.v_cache: torch.Size([1, 4, 114, 128]) + layer.3.k_cache: torch.Size([1, 4, 114, 128]) + layer.3.v_cache: torch.Size([1, 4, 114, 128]) + layer.4.k_cache: torch.Size([1, 4, 114, 128]) + layer.4.v_cache: torch.Size([1, 4, 114, 128]) + layer.4.output: torch.Size([1, 114, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.288s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 114, 128]) + layer.0.v_cache: torch.Size([1, 4, 114, 128]) + layer.1.k_cache: torch.Size([1, 4, 114, 128]) + layer.1.v_cache: torch.Size([1, 4, 114, 128]) + layer.2.k_cache: torch.Size([1, 4, 114, 128]) + layer.2.v_cache: torch.Size([1, 4, 114, 128]) + layer.3.k_cache: torch.Size([1, 4, 114, 128]) + layer.3.v_cache: torch.Size([1, 4, 114, 128]) + layer.4.k_cache: torch.Size([1, 4, 114, 128]) + layer.4.v_cache: torch.Size([1, 4, 114, 128]) + layer.4.output: torch.Size([1, 114, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10081583 26.68959019 + layer.0.v_cache 0.00001633 0.00485368 + layer.1.k_cache 0.08473034 3.29903961 + layer.1.v_cache 0.00000563 0.00173301 + layer.2.k_cache 0.00792928 0.45278375 + layer.2.v_cache 0.00001730 0.00453397 + layer.3.k_cache 0.01019665 2.01471777 + layer.3.v_cache 0.00001810 0.00505029 + layer.4.k_cache 0.00077074 0.10970189 + layer.4.v_cache 0.00005022 0.01073601 + layer.4.output 10.00037814 463.06641604 + ------------------------------------------------------------------------------------- + TOTAL 4.12983514 192.59162662 + (elements=992,256) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 992256 +Total Bytes 192016 +BPFP 1.5481 bits/point +EBPFP 3.0962 equivalent bits/point +MSE 192.591627 +---------------------- -------------------------------------------------------- +Time: 0.546s Load: 0.004s, Pack+Encode: 0.254s, Decode+Unpack: 0.288s +---------------------- -------------------------------------------------------- +💾 Converting with 192.5916 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-275.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-275.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-276.zst (34/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-276.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 98, 128) +Output shape: (1, 98, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.0.v_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.1.k_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.1.v_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.2.k_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.2.v_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.3.k_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.3.v_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.4.k_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.4.v_cache: torch.Size([1, 4, 98, 128]) -> torch.Size([1, 1, 98, 512]) + layer.4.output: torch.Size([1, 98, 3584]) -> torch.Size([1, 1, 98, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,364B, BPFP=0.8552 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,696B, BPFP=2.8214 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,688B, BPFP=1.5446 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 17,072B, BPFP=2.7219 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,036B, BPFP=1.7596 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,592B, BPFP=2.6454 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,512B, BPFP=1.6760 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,940B, BPFP=2.7009 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,952B, BPFP=2.3839 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,376B, BPFP=2.6110 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 36,792B, BPFP=0.8380 +⌛️ [2/4] FRONTEND: Frontend time: 0.235s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 98, 128]) + layer.0.v_cache: torch.Size([1, 4, 98, 128]) + layer.1.k_cache: torch.Size([1, 4, 98, 128]) + layer.1.v_cache: torch.Size([1, 4, 98, 128]) + layer.2.k_cache: torch.Size([1, 4, 98, 128]) + layer.2.v_cache: torch.Size([1, 4, 98, 128]) + layer.3.k_cache: torch.Size([1, 4, 98, 128]) + layer.3.v_cache: torch.Size([1, 4, 98, 128]) + layer.4.k_cache: torch.Size([1, 4, 98, 128]) + layer.4.v_cache: torch.Size([1, 4, 98, 128]) + layer.4.output: torch.Size([1, 98, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.324s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 98, 128]) + layer.0.v_cache: torch.Size([1, 4, 98, 128]) + layer.1.k_cache: torch.Size([1, 4, 98, 128]) + layer.1.v_cache: torch.Size([1, 4, 98, 128]) + layer.2.k_cache: torch.Size([1, 4, 98, 128]) + layer.2.v_cache: torch.Size([1, 4, 98, 128]) + layer.3.k_cache: torch.Size([1, 4, 98, 128]) + layer.3.v_cache: torch.Size([1, 4, 98, 128]) + layer.4.k_cache: torch.Size([1, 4, 98, 128]) + layer.4.v_cache: torch.Size([1, 4, 98, 128]) + layer.4.output: torch.Size([1, 98, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11008905 26.69701650 + layer.0.v_cache 0.00001718 0.00473273 + layer.1.k_cache 0.06240322 3.74613174 + layer.1.v_cache 0.00000548 0.00171537 + layer.2.k_cache 0.00473963 0.46531292 + layer.2.v_cache 0.00001718 0.00461231 + layer.3.k_cache 0.01141277 2.27681032 + layer.3.v_cache 0.00001769 0.00540670 + layer.4.k_cache 0.00076610 0.11697119 + layer.4.v_cache 0.00004814 0.01119249 + layer.4.output 0.02426045 548.61074162 + ------------------------------------------------------------------------------------- + TOTAL 0.02113762 227.85912315 + (elements=852,992) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 852992 +Total Bytes 173020 +BPFP 1.6227 bits/point +EBPFP 3.2454 equivalent bits/point +MSE 227.859123 +---------------------- -------------------------------------------------------- +Time: 0.564s Load: 0.004s, Pack+Encode: 0.235s, Decode+Unpack: 0.324s +---------------------- -------------------------------------------------------- +💾 Converting with 227.8591 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-276.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-276.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-280.zst (35/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-280.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 92, 128) +Output shape: (1, 92, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.output: torch.Size([1, 92, 3584]) -> torch.Size([1, 1, 92, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,560B, BPFP=0.9443 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,952B, BPFP=2.8791 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,732B, BPFP=1.6529 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,304B, BPFP=2.7690 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,972B, BPFP=1.8635 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,100B, BPFP=2.7344 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,636B, BPFP=1.8064 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,476B, BPFP=2.7982 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,388B, BPFP=2.4436 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,984B, BPFP=2.7147 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 42,576B, BPFP=1.0330 +⌛️ [2/4] FRONTEND: Frontend time: 0.184s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.250s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08919133 27.94896665 + layer.0.v_cache 0.00001414 0.00476348 + layer.1.k_cache 0.08591949 3.33880018 + layer.1.v_cache 0.00000528 0.00169925 + layer.2.k_cache 0.00389855 0.48418547 + layer.2.v_cache 0.00001738 0.00494193 + layer.3.k_cache 0.07679920 2.16713549 + layer.3.v_cache 0.00002028 0.00604713 + layer.4.k_cache 0.00072407 0.11474361 + layer.4.v_cache 0.00004779 0.01145035 + layer.4.output 0.14786680 577.50625970 + ------------------------------------------------------------------------------------- + TOTAL 0.07598265 239.80156185 + (elements=800,768) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 800768 +Total Bytes 175680 +BPFP 1.7551 bits/point +EBPFP 3.5102 equivalent bits/point +MSE 239.801562 +---------------------- -------------------------------------------------------- +Time: 0.438s Load: 0.004s, Pack+Encode: 0.184s, Decode+Unpack: 0.250s +---------------------- -------------------------------------------------------- +💾 Converting with 239.8016 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-280.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-280.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-284.zst (36/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-284.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 107, 128) +Output shape: (1, 107, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.0.v_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.1.k_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.1.v_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.2.k_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.2.v_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.3.k_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.3.v_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.4.k_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.4.v_cache: torch.Size([1, 4, 107, 128]) -> torch.Size([1, 1, 107, 512]) + layer.4.output: torch.Size([1, 107, 3584]) -> torch.Size([1, 1, 107, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,956B, BPFP=0.8697 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,464B, BPFP=2.8423 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,724B, BPFP=1.5660 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,856B, BPFP=2.7535 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 12,128B, BPFP=1.7710 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,452B, BPFP=2.6945 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,608B, BPFP=1.6951 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,812B, BPFP=2.7471 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 16,304B, BPFP=2.3808 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 18,260B, BPFP=2.6665 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 41,956B, BPFP=0.8753 +⌛️ [2/4] FRONTEND: Frontend time: 0.239s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 107, 128]) + layer.0.v_cache: torch.Size([1, 4, 107, 128]) + layer.1.k_cache: torch.Size([1, 4, 107, 128]) + layer.1.v_cache: torch.Size([1, 4, 107, 128]) + layer.2.k_cache: torch.Size([1, 4, 107, 128]) + layer.2.v_cache: torch.Size([1, 4, 107, 128]) + layer.3.k_cache: torch.Size([1, 4, 107, 128]) + layer.3.v_cache: torch.Size([1, 4, 107, 128]) + layer.4.k_cache: torch.Size([1, 4, 107, 128]) + layer.4.v_cache: torch.Size([1, 4, 107, 128]) + layer.4.output: torch.Size([1, 107, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.316s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 107, 128]) + layer.0.v_cache: torch.Size([1, 4, 107, 128]) + layer.1.k_cache: torch.Size([1, 4, 107, 128]) + layer.1.v_cache: torch.Size([1, 4, 107, 128]) + layer.2.k_cache: torch.Size([1, 4, 107, 128]) + layer.2.v_cache: torch.Size([1, 4, 107, 128]) + layer.3.k_cache: torch.Size([1, 4, 107, 128]) + layer.3.v_cache: torch.Size([1, 4, 107, 128]) + layer.4.k_cache: torch.Size([1, 4, 107, 128]) + layer.4.v_cache: torch.Size([1, 4, 107, 128]) + layer.4.output: torch.Size([1, 107, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11353527 26.87064198 + layer.0.v_cache 0.00001674 0.00489708 + layer.1.k_cache 0.04276301 2.99133301 + layer.1.v_cache 0.00000542 0.00169909 + layer.2.k_cache 0.00247095 0.46121850 + layer.2.v_cache 0.00001885 0.00474517 + layer.3.k_cache 0.02722364 2.17835699 + layer.3.v_cache 0.00001942 0.00514981 + layer.4.k_cache 0.00068256 0.11066856 + layer.4.v_cache 0.00004843 0.01095399 + layer.4.output 10.65454330 481.50016689 + ------------------------------------------------------------------------------------- + TOTAL 4.39815220 200.18475485 + (elements=931,328) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 931328 +Total Bytes 192520 +BPFP 1.6537 bits/point +EBPFP 3.3074 equivalent bits/point +MSE 200.184755 +---------------------- -------------------------------------------------------- +Time: 0.561s Load: 0.005s, Pack+Encode: 0.239s, Decode+Unpack: 0.316s +---------------------- -------------------------------------------------------- +💾 Converting with 200.1848 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-284.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-284.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-285.zst (37/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-285.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 84, 128) +Output shape: (1, 84, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.output: torch.Size([1, 84, 3584]) -> torch.Size([1, 1, 84, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,156B, BPFP=0.9591 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,044B, BPFP=2.9844 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,804B, BPFP=1.6376 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,252B, BPFP=2.8371 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,820B, BPFP=1.8266 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,992B, BPFP=2.7887 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,440B, BPFP=1.7560 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,452B, BPFP=2.8743 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,216B, BPFP=2.4583 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,976B, BPFP=2.7857 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,036B, BPFP=0.9044 +⌛️ [2/4] FRONTEND: Frontend time: 0.216s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.319s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12550225 27.40148344 + layer.0.v_cache 0.00001468 0.00507013 + layer.1.k_cache 0.05234316 3.56451779 + layer.1.v_cache 0.00000508 0.00161344 + layer.2.k_cache 0.00733058 0.42316164 + layer.2.v_cache 0.00001689 0.00466149 + layer.3.k_cache 0.01592888 2.27490016 + layer.3.v_cache 0.00001847 0.00551486 + layer.4.k_cache 0.00069460 0.10986504 + layer.4.v_cache 0.00005205 0.01084297 + layer.4.output 0.16184913 621.10039328 + ------------------------------------------------------------------------------------- + TOTAL 0.07852062 257.73555200 + (elements=731,136) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 731136 +Total Bytes 157188 +BPFP 1.7199 bits/point +EBPFP 3.4399 equivalent bits/point +MSE 257.735552 +---------------------- -------------------------------------------------------- +Time: 0.539s Load: 0.004s, Pack+Encode: 0.216s, Decode+Unpack: 0.319s +---------------------- -------------------------------------------------------- +💾 Converting with 257.7356 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-285.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-285.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-293.zst (38/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-293.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 78, 128) +Output shape: (1, 78, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) -> torch.Size([1, 1, 78, 512]) + layer.4.output: torch.Size([1, 78, 3584]) -> torch.Size([1, 1, 78, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,972B, BPFP=0.9960 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,464B, BPFP=3.0978 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,380B, BPFP=1.6787 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,672B, BPFP=2.9391 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,596B, BPFP=1.9223 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,052B, BPFP=2.8149 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,888B, BPFP=1.7804 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,584B, BPFP=2.9215 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,772B, BPFP=2.5585 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,264B, BPFP=2.8574 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,352B, BPFP=0.8972 +⌛️ [2/4] FRONTEND: Frontend time: 0.269s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.output: torch.Size([1, 78, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.326s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 78, 128]) + layer.0.v_cache: torch.Size([1, 4, 78, 128]) + layer.1.k_cache: torch.Size([1, 4, 78, 128]) + layer.1.v_cache: torch.Size([1, 4, 78, 128]) + layer.2.k_cache: torch.Size([1, 4, 78, 128]) + layer.2.v_cache: torch.Size([1, 4, 78, 128]) + layer.3.k_cache: torch.Size([1, 4, 78, 128]) + layer.3.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.k_cache: torch.Size([1, 4, 78, 128]) + layer.4.v_cache: torch.Size([1, 4, 78, 128]) + layer.4.output: torch.Size([1, 78, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11879376 32.10539050 + layer.0.v_cache 0.00001438 0.00493962 + layer.1.k_cache 0.05518859 3.46388440 + layer.1.v_cache 0.00000514 0.00179781 + layer.2.k_cache 0.00243519 0.47345235 + layer.2.v_cache 0.00001742 0.00484607 + layer.3.k_cache 0.07959401 2.56699469 + layer.3.v_cache 0.00001812 0.00586675 + layer.4.k_cache 0.00069358 0.11656530 + layer.4.v_cache 0.00004620 0.01155311 + layer.4.output 0.17426339 689.40436126 + ------------------------------------------------------------------------------------- + TOTAL 0.08686177 286.15210703 + (elements=678,912) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 678912 +Total Bytes 148996 +BPFP 1.7557 bits/point +EBPFP 3.5114 equivalent bits/point +MSE 286.152107 +---------------------- -------------------------------------------------------- +Time: 0.600s Load: 0.005s, Pack+Encode: 0.269s, Decode+Unpack: 0.326s +---------------------- -------------------------------------------------------- +💾 Converting with 286.1521 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-293.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-293.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-299.zst (39/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-299.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 82, 128) +Output shape: (1, 82, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.0.v_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.1.k_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.1.v_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.2.k_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.2.v_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.3.k_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.3.v_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.4.k_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.4.v_cache: torch.Size([1, 4, 82, 128]) -> torch.Size([1, 1, 82, 512]) + layer.4.output: torch.Size([1, 82, 3584]) -> torch.Size([1, 1, 82, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,080B, BPFP=0.9680 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,792B, BPFP=3.0091 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,748B, BPFP=1.6669 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,268B, BPFP=2.9093 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,856B, BPFP=1.8780 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,784B, BPFP=2.8171 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,348B, BPFP=1.7812 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,164B, BPFP=2.8895 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,124B, BPFP=2.5008 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,852B, BPFP=2.8300 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,112B, BPFP=0.8741 +⌛️ [2/4] FRONTEND: Frontend time: 0.269s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 82, 128]) + layer.0.v_cache: torch.Size([1, 4, 82, 128]) + layer.1.k_cache: torch.Size([1, 4, 82, 128]) + layer.1.v_cache: torch.Size([1, 4, 82, 128]) + layer.2.k_cache: torch.Size([1, 4, 82, 128]) + layer.2.v_cache: torch.Size([1, 4, 82, 128]) + layer.3.k_cache: torch.Size([1, 4, 82, 128]) + layer.3.v_cache: torch.Size([1, 4, 82, 128]) + layer.4.k_cache: torch.Size([1, 4, 82, 128]) + layer.4.v_cache: torch.Size([1, 4, 82, 128]) + layer.4.output: torch.Size([1, 82, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.323s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 82, 128]) + layer.0.v_cache: torch.Size([1, 4, 82, 128]) + layer.1.k_cache: torch.Size([1, 4, 82, 128]) + layer.1.v_cache: torch.Size([1, 4, 82, 128]) + layer.2.k_cache: torch.Size([1, 4, 82, 128]) + layer.2.v_cache: torch.Size([1, 4, 82, 128]) + layer.3.k_cache: torch.Size([1, 4, 82, 128]) + layer.3.v_cache: torch.Size([1, 4, 82, 128]) + layer.4.k_cache: torch.Size([1, 4, 82, 128]) + layer.4.v_cache: torch.Size([1, 4, 82, 128]) + layer.4.output: torch.Size([1, 82, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11325797 29.88421184 + layer.0.v_cache 0.00001565 0.00504233 + layer.1.k_cache 0.07651102 3.27263734 + layer.1.v_cache 0.00000530 0.00186219 + layer.2.k_cache 0.00872486 0.49174779 + layer.2.v_cache 0.00001696 0.00500643 + layer.3.k_cache 0.12876377 2.36573270 + layer.3.v_cache 0.00001905 0.00612663 + layer.4.k_cache 0.00069190 0.11570972 + layer.4.v_cache 0.00004968 0.01211293 + layer.4.output 0.16578155 648.02161368 + ------------------------------------------------------------------------------------- + TOTAL 0.08756041 268.95949916 + (elements=713,728) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 713728 +Total Bytes 154128 +BPFP 1.7276 bits/point +EBPFP 3.4552 equivalent bits/point +MSE 268.959499 +---------------------- -------------------------------------------------------- +Time: 0.597s Load: 0.005s, Pack+Encode: 0.269s, Decode+Unpack: 0.323s +---------------------- -------------------------------------------------------- +💾 Converting with 268.9595 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-299.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-299.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-301.zst (40/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-301.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,020B, BPFP=1.0187 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,084B, BPFP=3.0609 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,500B, BPFP=1.7248 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,620B, BPFP=2.9667 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,584B, BPFP=1.9448 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,068B, BPFP=2.8547 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,932B, BPFP=1.8125 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,588B, BPFP=2.9602 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,652B, BPFP=2.5674 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,180B, BPFP=2.8774 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,688B, BPFP=0.8896 +⌛️ [2/4] FRONTEND: Frontend time: 0.270s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.291s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12835005 31.61633142 + layer.0.v_cache 0.00001616 0.00524046 + layer.1.k_cache 0.05672838 3.51262159 + layer.1.v_cache 0.00000535 0.00225246 + layer.2.k_cache 0.00240846 0.53407605 + layer.2.v_cache 0.00001644 0.00514888 + layer.3.k_cache 0.04570918 2.59645972 + layer.3.v_cache 0.00001733 0.00624639 + layer.4.k_cache 0.00072090 0.12261111 + layer.4.v_cache 0.00004908 0.01259731 + layer.4.output 0.18444509 697.64506030 + ------------------------------------------------------------------------------------- + TOTAL 0.08971394 289.52523573 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 147916 +BPFP 1.7656 bits/point +EBPFP 3.5312 equivalent bits/point +MSE 289.525236 +---------------------- -------------------------------------------------------- +Time: 0.566s Load: 0.005s, Pack+Encode: 0.270s, Decode+Unpack: 0.291s +---------------------- -------------------------------------------------------- +💾 Converting with 289.5252 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-301.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-301.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-313.zst (41/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-313.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 71, 128) +Output shape: (1, 71, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.output: torch.Size([1, 71, 3584]) -> torch.Size([1, 1, 71, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,780B, BPFP=1.0519 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,508B, BPFP=3.1928 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,112B, BPFP=1.7852 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,428B, BPFP=2.9551 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,952B, BPFP=1.9701 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,680B, BPFP=2.7905 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,288B, BPFP=1.8239 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,192B, BPFP=2.9032 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,864B, BPFP=2.6109 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,252B, BPFP=2.9164 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,992B, BPFP=0.9429 +⌛️ [2/4] FRONTEND: Frontend time: 0.246s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.267s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10085608 29.68769944 + layer.0.v_cache 0.00001554 0.00462977 + layer.1.k_cache 0.06589249 3.50661845 + layer.1.v_cache 0.00000499 0.00176464 + layer.2.k_cache 0.00422055 0.49438965 + layer.2.v_cache 0.00001579 0.00437542 + layer.3.k_cache 0.05337993 2.43404077 + layer.3.v_cache 0.00001747 0.00573243 + layer.4.k_cache 0.00075412 0.11813354 + layer.4.v_cache 0.00004712 0.01150151 + layer.4.output 0.19135183 751.68202968 + ------------------------------------------------------------------------------------- + TOTAL 0.09203923 311.64959373 + (elements=617,984) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 617984 +Total Bytes 139048 +BPFP 1.8000 bits/point +EBPFP 3.6000 equivalent bits/point +MSE 311.649594 +---------------------- -------------------------------------------------------- +Time: 0.515s Load: 0.003s, Pack+Encode: 0.246s, Decode+Unpack: 0.267s +---------------------- -------------------------------------------------------- +💾 Converting with 311.6496 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-313.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-313.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-315.zst (42/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-315.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 84, 128) +Output shape: (1, 84, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.output: torch.Size([1, 84, 3584]) -> torch.Size([1, 1, 84, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,192B, BPFP=0.9658 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,600B, BPFP=2.9018 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,860B, BPFP=1.6481 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,028B, BPFP=2.7954 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,956B, BPFP=1.8519 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,616B, BPFP=2.7188 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,476B, BPFP=1.7626 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,164B, BPFP=2.8207 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,124B, BPFP=2.4412 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,632B, BPFP=2.7217 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,568B, BPFP=0.8654 +⌛️ [2/4] FRONTEND: Frontend time: 0.255s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.287s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09057366 28.23681641 + layer.0.v_cache 0.00001789 0.00497122 + layer.1.k_cache 0.03614600 3.40125965 + layer.1.v_cache 0.00000481 0.00174813 + layer.2.k_cache 0.00554357 0.48403159 + layer.2.v_cache 0.00001714 0.00458462 + layer.3.k_cache 0.02742965 2.25141543 + layer.3.v_cache 0.00001729 0.00572267 + layer.4.k_cache 0.00093027 0.11310725 + layer.4.v_cache 0.00004906 0.01110303 + layer.4.output 0.16441821 627.50786565 + ------------------------------------------------------------------------------------- + TOTAL 0.07715628 260.41587174 + (elements=731,136) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 731136 +Total Bytes 154216 +BPFP 1.6874 bits/point +EBPFP 3.3748 equivalent bits/point +MSE 260.415872 +---------------------- -------------------------------------------------------- +Time: 0.545s Load: 0.003s, Pack+Encode: 0.255s, Decode+Unpack: 0.287s +---------------------- -------------------------------------------------------- +💾 Converting with 260.4159 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-315.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-315.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-316.zst (43/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-316.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,056B, BPFP=1.0260 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,284B, BPFP=3.1015 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,532B, BPFP=1.7313 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,568B, BPFP=2.9562 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,572B, BPFP=1.9424 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,048B, BPFP=2.8506 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,916B, BPFP=1.8093 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,772B, BPFP=2.9976 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,688B, BPFP=2.5747 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,248B, BPFP=2.8912 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,472B, BPFP=0.9123 +⌛️ [2/4] FRONTEND: Frontend time: 0.279s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.290s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09325875 30.55671672 + layer.0.v_cache 0.00001406 0.00513163 + layer.1.k_cache 0.05775181 3.48878974 + layer.1.v_cache 0.00000524 0.00187233 + layer.2.k_cache 0.00413042 0.53868727 + layer.2.v_cache 0.00001743 0.00496489 + layer.3.k_cache 0.04387648 2.45412356 + layer.3.v_cache 0.00002497 0.00655067 + layer.4.k_cache 0.00071512 0.12704193 + layer.4.v_cache 0.00005147 0.01280951 + layer.4.output 0.17652170 697.64221939 + ------------------------------------------------------------------------------------- + TOTAL 0.08444104 289.45248376 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 149156 +BPFP 1.7804 bits/point +EBPFP 3.5608 equivalent bits/point +MSE 289.452484 +---------------------- -------------------------------------------------------- +Time: 0.574s Load: 0.005s, Pack+Encode: 0.279s, Decode+Unpack: 0.290s +---------------------- -------------------------------------------------------- +💾 Converting with 289.4525 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-316.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-316.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-324.zst (44/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-324.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 88, 128) +Output shape: (1, 88, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.output: torch.Size([1, 88, 3584]) -> torch.Size([1, 1, 88, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,396B, BPFP=0.9581 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,320B, BPFP=2.7202 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,988B, BPFP=1.5959 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,800B, BPFP=2.6278 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,896B, BPFP=1.7571 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,252B, BPFP=2.5305 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,268B, BPFP=1.6456 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,900B, BPFP=2.6456 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,028B, BPFP=2.3132 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,372B, BPFP=2.5518 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,324B, BPFP=0.7438 +⌛️ [2/4] FRONTEND: Frontend time: 0.240s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.278s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08212358 27.98851152 + layer.0.v_cache 0.00001682 0.00412047 + layer.1.k_cache 0.05918076 3.48936393 + layer.1.v_cache 0.00000473 0.00144546 + layer.2.k_cache 0.00260097 0.44645639 + layer.2.v_cache 0.00001544 0.00418609 + layer.3.k_cache 0.02838851 2.05642301 + layer.3.v_cache 0.00001657 0.00516545 + layer.4.k_cache 0.00106615 0.10948386 + layer.4.v_cache 0.00004456 0.01007707 + layer.4.output 0.17173813 602.06275365 + ------------------------------------------------------------------------------------- + TOTAL 0.08091912 249.91497111 + (elements=765,952) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 765952 +Total Bytes 149544 +BPFP 1.5619 bits/point +EBPFP 3.1238 equivalent bits/point +MSE 249.914971 +---------------------- -------------------------------------------------------- +Time: 0.522s Load: 0.005s, Pack+Encode: 0.240s, Decode+Unpack: 0.278s +---------------------- -------------------------------------------------------- +💾 Converting with 249.9150 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-324.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-324.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-325.zst (45/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-325.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 88, 128) +Output shape: (1, 88, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.output: torch.Size([1, 88, 3584]) -> torch.Size([1, 1, 88, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,348B, BPFP=0.9496 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,972B, BPFP=2.8359 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,204B, BPFP=1.6342 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,500B, BPFP=2.7521 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,220B, BPFP=1.8146 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,044B, BPFP=2.6712 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,584B, BPFP=1.7017 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,512B, BPFP=2.7543 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,636B, BPFP=2.4212 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,096B, BPFP=2.6804 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,464B, BPFP=0.8742 +⌛️ [2/4] FRONTEND: Frontend time: 0.270s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.309s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10731998 28.40448275 + layer.0.v_cache 0.00001618 0.00458869 + layer.1.k_cache 0.07375650 3.33109214 + layer.1.v_cache 0.00000540 0.00169113 + layer.2.k_cache 0.00246326 0.45111370 + layer.2.v_cache 0.00001746 0.00483207 + layer.3.k_cache 0.01642703 1.86339170 + layer.3.v_cache 0.00001778 0.00545736 + layer.4.k_cache 0.00066607 0.11148694 + layer.4.v_cache 0.00004728 0.01115044 + layer.4.output 0.15452130 601.08431412 + ------------------------------------------------------------------------------------- + TOTAL 0.07543447 249.51644034 + (elements=765,952) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 765952 +Total Bytes 159580 +BPFP 1.6667 bits/point +EBPFP 3.3335 equivalent bits/point +MSE 249.516440 +---------------------- -------------------------------------------------------- +Time: 0.583s Load: 0.004s, Pack+Encode: 0.270s, Decode+Unpack: 0.309s +---------------------- -------------------------------------------------------- +💾 Converting with 249.5164 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-325.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-325.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-328.zst (46/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-328.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 99, 128) +Output shape: (1, 99, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.0.v_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.1.k_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.1.v_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.2.k_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.2.v_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.3.k_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.3.v_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.4.k_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.4.v_cache: torch.Size([1, 4, 99, 128]) -> torch.Size([1, 1, 99, 512]) + layer.4.output: torch.Size([1, 99, 3584]) -> torch.Size([1, 1, 99, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,804B, BPFP=0.9160 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,940B, BPFP=2.8314 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,880B, BPFP=1.5593 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 17,372B, BPFP=2.7418 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,240B, BPFP=1.7740 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,984B, BPFP=2.6806 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,632B, BPFP=1.6780 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 17,384B, BPFP=2.7437 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,944B, BPFP=2.3586 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,812B, BPFP=2.6534 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 39,508B, BPFP=0.8908 +⌛️ [2/4] FRONTEND: Frontend time: 0.227s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 99, 128]) + layer.0.v_cache: torch.Size([1, 4, 99, 128]) + layer.1.k_cache: torch.Size([1, 4, 99, 128]) + layer.1.v_cache: torch.Size([1, 4, 99, 128]) + layer.2.k_cache: torch.Size([1, 4, 99, 128]) + layer.2.v_cache: torch.Size([1, 4, 99, 128]) + layer.3.k_cache: torch.Size([1, 4, 99, 128]) + layer.3.v_cache: torch.Size([1, 4, 99, 128]) + layer.4.k_cache: torch.Size([1, 4, 99, 128]) + layer.4.v_cache: torch.Size([1, 4, 99, 128]) + layer.4.output: torch.Size([1, 99, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.285s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 99, 128]) + layer.0.v_cache: torch.Size([1, 4, 99, 128]) + layer.1.k_cache: torch.Size([1, 4, 99, 128]) + layer.1.v_cache: torch.Size([1, 4, 99, 128]) + layer.2.k_cache: torch.Size([1, 4, 99, 128]) + layer.2.v_cache: torch.Size([1, 4, 99, 128]) + layer.3.k_cache: torch.Size([1, 4, 99, 128]) + layer.3.v_cache: torch.Size([1, 4, 99, 128]) + layer.4.k_cache: torch.Size([1, 4, 99, 128]) + layer.4.v_cache: torch.Size([1, 4, 99, 128]) + layer.4.output: torch.Size([1, 99, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09061131 26.52684560 + layer.0.v_cache 0.00001461 0.00435008 + layer.1.k_cache 0.05062923 3.25675363 + layer.1.v_cache 0.00000709 0.00169401 + layer.2.k_cache 0.00392854 0.44544829 + layer.2.v_cache 0.00001793 0.00434839 + layer.3.k_cache 0.05082532 2.17054456 + layer.3.v_cache 0.00001820 0.00515636 + layer.4.k_cache 0.00080398 0.10855463 + layer.4.v_cache 0.00005053 0.00997472 + layer.4.output 0.02403493 541.40043290 + ------------------------------------------------------------------------------------- + TOTAL 0.02147948 224.84333533 + (elements=861,696) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 861696 +Total Bytes 178500 +BPFP 1.6572 bits/point +EBPFP 3.3144 equivalent bits/point +MSE 224.843335 +---------------------- -------------------------------------------------------- +Time: 0.517s Load: 0.005s, Pack+Encode: 0.227s, Decode+Unpack: 0.285s +---------------------- -------------------------------------------------------- +💾 Converting with 224.8433 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-328.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-328.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-33.zst (47/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-33.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 71, 128) +Output shape: (1, 71, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.output: torch.Size([1, 71, 3584]) -> torch.Size([1, 1, 71, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,788B, BPFP=1.0537 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,720B, BPFP=3.2394 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,364B, BPFP=1.8407 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,800B, BPFP=3.0370 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,432B, BPFP=2.0757 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,388B, BPFP=2.9463 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,760B, BPFP=1.9278 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,304B, BPFP=3.1479 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,016B, BPFP=2.6444 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,596B, BPFP=2.9921 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,372B, BPFP=1.0177 +⌛️ [2/4] FRONTEND: Frontend time: 0.248s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.333s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11136492 30.13230359 + layer.0.v_cache 0.00001955 0.00516209 + layer.1.k_cache 0.06280328 3.50826811 + layer.1.v_cache 0.00000544 0.00191268 + layer.2.k_cache 0.00235174 0.51208179 + layer.2.v_cache 0.00001859 0.00517673 + layer.3.k_cache 0.03326307 2.30550933 + layer.3.v_cache 0.00001926 0.00598906 + layer.4.k_cache 0.00064473 0.11571959 + layer.4.v_cache 0.00005254 0.01154232 + layer.4.output 0.19494371 742.58475855 + ------------------------------------------------------------------------------------- + TOTAL 0.09265583 307.92335148 + (elements=617,984) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 617984 +Total Bytes 145540 +BPFP 1.8841 bits/point +EBPFP 3.7681 equivalent bits/point +MSE 307.923351 +---------------------- -------------------------------------------------------- +Time: 0.584s Load: 0.003s, Pack+Encode: 0.248s, Decode+Unpack: 0.333s +---------------------- -------------------------------------------------------- +💾 Converting with 307.9234 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-33.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-33.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-332.zst (48/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-332.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 74, 128) +Output shape: (1, 74, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.output: torch.Size([1, 74, 3584]) -> torch.Size([1, 1, 74, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,012B, BPFP=1.0583 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,188B, BPFP=3.2069 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,368B, BPFP=1.7669 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,152B, BPFP=2.9882 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,792B, BPFP=2.0676 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,848B, BPFP=2.9240 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,968B, BPFP=1.8936 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,648B, BPFP=3.0929 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,484B, BPFP=2.6360 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,300B, BPFP=3.0194 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,948B, BPFP=0.9335 +⌛️ [2/4] FRONTEND: Frontend time: 0.250s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.313s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08829598 30.97784589 + layer.0.v_cache 0.00001455 0.00551398 + layer.1.k_cache 0.05950350 3.37114612 + layer.1.v_cache 0.00000529 0.00197730 + layer.2.k_cache 0.00411868 0.50223227 + layer.2.v_cache 0.00001680 0.00488780 + layer.3.k_cache 0.04877868 2.39256245 + layer.3.v_cache 0.00001896 0.00661326 + layer.4.k_cache 0.00063741 0.12264716 + layer.4.v_cache 0.00004764 0.01229914 + layer.4.output 0.18365649 710.40842181 + ------------------------------------------------------------------------------------- + TOTAL 0.08747253 294.72098106 + (elements=644,096) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 644096 +Total Bytes 147708 +BPFP 1.8346 bits/point +EBPFP 3.6692 equivalent bits/point +MSE 294.720981 +---------------------- -------------------------------------------------------- +Time: 0.568s Load: 0.005s, Pack+Encode: 0.250s, Decode+Unpack: 0.313s +---------------------- -------------------------------------------------------- +💾 Converting with 294.7210 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-332.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-332.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-336.zst (49/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-336.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.007s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,956B, BPFP=1.0057 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,336B, BPFP=3.1120 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,600B, BPFP=1.7451 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,796B, BPFP=3.0024 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,724B, BPFP=1.9732 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,396B, BPFP=2.9213 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,964B, BPFP=1.8190 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,752B, BPFP=2.9935 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,852B, BPFP=2.6080 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,480B, BPFP=2.9383 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,604B, BPFP=0.8872 +⌛️ [2/4] FRONTEND: Frontend time: 0.258s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.292s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11195674 30.86276443 + layer.0.v_cache 0.00001346 0.00489195 + layer.1.k_cache 0.03755230 3.59753814 + layer.1.v_cache 0.00000535 0.00195009 + layer.2.k_cache 0.00239600 0.52287332 + layer.2.v_cache 0.00001753 0.00535250 + layer.3.k_cache 0.02867049 2.39066444 + layer.3.v_cache 0.00001751 0.00592374 + layer.4.k_cache 0.00071811 0.12470361 + layer.4.v_cache 0.00005486 0.01257418 + layer.4.output 0.19495771 698.89737941 + ------------------------------------------------------------------------------------- + TOTAL 0.09094743 289.98887601 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 149460 +BPFP 1.7840 bits/point +EBPFP 3.5681 equivalent bits/point +MSE 289.988876 +---------------------- -------------------------------------------------------- +Time: 0.557s Load: 0.007s, Pack+Encode: 0.258s, Decode+Unpack: 0.292s +---------------------- -------------------------------------------------------- +💾 Converting with 289.9889 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-336.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-336.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-342.zst (50/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-342.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 90, 128) +Output shape: (1, 90, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) -> torch.Size([1, 1, 90, 512]) + layer.4.output: torch.Size([1, 90, 3584]) -> torch.Size([1, 1, 90, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,452B, BPFP=0.9465 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,188B, BPFP=2.8104 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,308B, BPFP=1.6160 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,632B, BPFP=2.7139 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,352B, BPFP=1.7972 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,264B, BPFP=2.6500 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,860B, BPFP=1.7118 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,772B, BPFP=2.7382 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,920B, BPFP=2.4167 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,384B, BPFP=2.6708 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,156B, BPFP=0.7975 +⌛️ [2/4] FRONTEND: Frontend time: 0.242s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.299s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 90, 128]) + layer.0.v_cache: torch.Size([1, 4, 90, 128]) + layer.1.k_cache: torch.Size([1, 4, 90, 128]) + layer.1.v_cache: torch.Size([1, 4, 90, 128]) + layer.2.k_cache: torch.Size([1, 4, 90, 128]) + layer.2.v_cache: torch.Size([1, 4, 90, 128]) + layer.3.k_cache: torch.Size([1, 4, 90, 128]) + layer.3.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.k_cache: torch.Size([1, 4, 90, 128]) + layer.4.v_cache: torch.Size([1, 4, 90, 128]) + layer.4.output: torch.Size([1, 90, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08657128 29.66180284 + layer.0.v_cache 0.00001574 0.00500133 + layer.1.k_cache 0.05618976 3.31095683 + layer.1.v_cache 0.00000506 0.00172484 + layer.2.k_cache 0.00392605 0.44029244 + layer.2.v_cache 0.00001570 0.00452096 + layer.3.k_cache 0.02849737 2.30185547 + layer.3.v_cache 0.00001769 0.00629981 + layer.4.k_cache 0.00099738 0.11312606 + layer.4.v_cache 0.00004751 0.01213734 + layer.4.output 0.15107292 571.51329365 + ------------------------------------------------------------------------------------- + TOTAL 0.07257612 237.43828079 + (elements=783,360) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 783360 +Total Bytes 159288 +BPFP 1.6267 bits/point +EBPFP 3.2534 equivalent bits/point +MSE 237.438281 +---------------------- -------------------------------------------------------- +Time: 0.545s Load: 0.004s, Pack+Encode: 0.242s, Decode+Unpack: 0.299s +---------------------- -------------------------------------------------------- +💾 Converting with 237.4383 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-342.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-342.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-346.zst (51/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-346.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 80, 128) +Output shape: (1, 80, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.output: torch.Size([1, 80, 3584]) -> torch.Size([1, 1, 80, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,968B, BPFP=0.9703 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,220B, BPFP=2.9727 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,544B, BPFP=1.6687 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,600B, BPFP=2.8516 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,700B, BPFP=1.8945 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,288B, BPFP=2.7906 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,060B, BPFP=1.7695 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,772B, BPFP=2.8852 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,864B, BPFP=2.5125 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,440B, BPFP=2.8203 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,292B, BPFP=0.8452 +⌛️ [2/4] FRONTEND: Frontend time: 0.261s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.290s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09254595 28.92623291 + layer.0.v_cache 0.00001776 0.00470066 + layer.1.k_cache 0.05775445 3.41446152 + layer.1.v_cache 0.00000504 0.00171463 + layer.2.k_cache 0.00891839 0.46332550 + layer.2.v_cache 0.00001732 0.00506927 + layer.3.k_cache 0.02836140 2.22389297 + layer.3.v_cache 0.00001672 0.00582427 + layer.4.k_cache 0.00083566 0.10952775 + layer.4.v_cache 0.00004876 0.01180479 + layer.4.output 0.16993865 656.34263393 + ------------------------------------------------------------------------------------- + TOTAL 0.08106423 272.32735246 + (elements=696,320) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 696320 +Total Bytes 148748 +BPFP 1.7090 bits/point +EBPFP 3.4179 equivalent bits/point +MSE 272.327352 +---------------------- -------------------------------------------------------- +Time: 0.554s Load: 0.004s, Pack+Encode: 0.261s, Decode+Unpack: 0.290s +---------------------- -------------------------------------------------------- +💾 Converting with 272.3274 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-346.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-346.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-348.zst (52/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-348.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 97, 128) +Output shape: (1, 97, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.output: torch.Size([1, 97, 3584]) -> torch.Size([1, 1, 97, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,244B, BPFP=0.8447 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,644B, BPFP=2.6811 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,376B, BPFP=1.5103 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,968B, BPFP=2.5722 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,116B, BPFP=1.6295 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,560B, BPFP=2.5064 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,800B, BPFP=1.5786 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,060B, BPFP=2.5870 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,920B, BPFP=2.2423 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,632B, BPFP=2.5180 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,848B, BPFP=0.7789 +⌛️ [2/4] FRONTEND: Frontend time: 0.243s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.output: torch.Size([1, 97, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.288s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.output: torch.Size([1, 97, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13093194 29.89068541 + layer.0.v_cache 0.00001635 0.00504953 + layer.1.k_cache 0.05235285 3.48651815 + layer.1.v_cache 0.00000496 0.00179754 + layer.2.k_cache 0.00512381 0.45207832 + layer.2.v_cache 0.00001647 0.00463233 + layer.3.k_cache 0.01138228 2.06428371 + layer.3.v_cache 0.00001850 0.00556022 + layer.4.k_cache 0.00081880 0.10935148 + layer.4.v_cache 0.00004796 0.01004488 + layer.4.output 0.02447439 563.27328792 + ------------------------------------------------------------------------------------- + TOTAL 0.02188439 234.05547159 + (elements=844,288) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 844288 +Total Bytes 162168 +BPFP 1.5366 bits/point +EBPFP 3.0732 equivalent bits/point +MSE 234.055472 +---------------------- -------------------------------------------------------- +Time: 0.535s Load: 0.004s, Pack+Encode: 0.243s, Decode+Unpack: 0.288s +---------------------- -------------------------------------------------------- +💾 Converting with 234.0555 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-348.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-348.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-356.zst (53/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-356.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 79, 128) +Output shape: (1, 79, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.output: torch.Size([1, 79, 3584]) -> torch.Size([1, 1, 79, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,932B, BPFP=0.9755 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,712B, BPFP=3.1076 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,428B, BPFP=1.6669 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,776B, BPFP=2.9225 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,772B, BPFP=1.9328 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,552B, BPFP=2.8782 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,980B, BPFP=1.7761 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,688B, BPFP=2.9051 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,768B, BPFP=2.5253 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,296B, BPFP=2.8275 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,080B, BPFP=0.9064 +⌛️ [2/4] FRONTEND: Frontend time: 0.229s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.output: torch.Size([1, 79, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.282s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.output: torch.Size([1, 79, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09510638 31.20119660 + layer.0.v_cache 0.00001399 0.00476386 + layer.1.k_cache 0.05467150 3.49095250 + layer.1.v_cache 0.00000511 0.00178445 + layer.2.k_cache 0.00545186 0.44744796 + layer.2.v_cache 0.00001652 0.00486989 + layer.3.k_cache 0.02986256 2.13666235 + layer.3.v_cache 0.00001747 0.00503519 + layer.4.k_cache 0.00068266 0.11054954 + layer.4.v_cache 0.00005277 0.01128566 + layer.4.output 0.17642128 660.92540687 + ------------------------------------------------------------------------------------- + TOTAL 0.08357822 274.34661154 + (elements=687,616) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 687616 +Total Bytes 150984 +BPFP 1.7566 bits/point +EBPFP 3.5132 equivalent bits/point +MSE 274.346612 +---------------------- -------------------------------------------------------- +Time: 0.515s Load: 0.004s, Pack+Encode: 0.229s, Decode+Unpack: 0.282s +---------------------- -------------------------------------------------------- +💾 Converting with 274.3466 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-356.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-356.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-377.zst (54/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-377.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,020B, BPFP=1.0187 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,216B, BPFP=3.0877 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,552B, BPFP=1.7354 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,676B, BPFP=2.9781 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,636B, BPFP=1.9554 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,140B, BPFP=2.8693 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,980B, BPFP=1.8222 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,712B, BPFP=2.9854 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,708B, BPFP=2.5787 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,432B, BPFP=2.9286 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,656B, BPFP=0.8887 +⌛️ [2/4] FRONTEND: Frontend time: 0.246s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.277s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08730992 31.28802570 + layer.0.v_cache 0.00001543 0.00535935 + layer.1.k_cache 0.05635426 3.56024646 + layer.1.v_cache 0.00000520 0.00202881 + layer.2.k_cache 0.00390536 0.49138042 + layer.2.v_cache 0.00001661 0.00528406 + layer.3.k_cache 0.06483487 2.43511527 + layer.3.v_cache 0.00002010 0.00618051 + layer.4.k_cache 0.00072652 0.12063687 + layer.4.v_cache 0.00004842 0.01209608 + layer.4.output 0.18845482 697.26223330 + ------------------------------------------------------------------------------------- + TOTAL 0.09014238 289.33894039 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 148728 +BPFP 1.7753 bits/point +EBPFP 3.5506 equivalent bits/point +MSE 289.338940 +---------------------- -------------------------------------------------------- +Time: 0.526s Load: 0.004s, Pack+Encode: 0.246s, Decode+Unpack: 0.277s +---------------------- -------------------------------------------------------- +💾 Converting with 289.3389 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-377.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-377.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-383.zst (55/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-383.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 68, 128) +Output shape: (1, 68, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.output: torch.Size([1, 68, 3584]) -> torch.Size([1, 1, 68, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,592B, BPFP=1.0551 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,220B, BPFP=3.2675 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,048B, BPFP=1.8493 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,044B, BPFP=2.9972 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,888B, BPFP=2.0423 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,676B, BPFP=2.9127 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,212B, BPFP=1.8869 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,412B, BPFP=3.0818 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,704B, BPFP=2.6893 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,276B, BPFP=3.0506 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,660B, BPFP=1.0393 +⌛️ [2/4] FRONTEND: Frontend time: 0.221s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.266s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10784986 31.86296530 + layer.0.v_cache 0.00001387 0.00458919 + layer.1.k_cache 0.01409798 3.57679345 + layer.1.v_cache 0.00000489 0.00160370 + layer.2.k_cache 0.00254294 0.45983368 + layer.2.v_cache 0.00001653 0.00468769 + layer.3.k_cache 0.05709259 2.04246027 + layer.3.v_cache 0.00001736 0.00536741 + layer.4.k_cache 0.00062715 0.10918020 + layer.4.v_cache 0.00004758 0.01095532 + layer.4.output 0.22376562 768.16576943 + ------------------------------------------------------------------------------------- + TOTAL 0.10286295 318.54346013 + (elements=591,872) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 591872 +Total Bytes 139732 +BPFP 1.8887 bits/point +EBPFP 3.7774 equivalent bits/point +MSE 318.543460 +---------------------- -------------------------------------------------------- +Time: 0.490s Load: 0.003s, Pack+Encode: 0.221s, Decode+Unpack: 0.266s +---------------------- -------------------------------------------------------- +💾 Converting with 318.5435 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-383.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-383.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-386.zst (56/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-386.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 79, 128) +Output shape: (1, 79, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) -> torch.Size([1, 1, 79, 512]) + layer.4.output: torch.Size([1, 79, 3584]) -> torch.Size([1, 1, 79, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,000B, BPFP=0.9889 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,996B, BPFP=2.9660 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,480B, BPFP=1.6772 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,252B, BPFP=2.8188 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,628B, BPFP=1.9043 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,960B, BPFP=2.7611 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,860B, BPFP=1.7524 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,432B, BPFP=2.8544 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,620B, BPFP=2.4960 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,116B, BPFP=2.7919 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,144B, BPFP=0.9082 +⌛️ [2/4] FRONTEND: Frontend time: 0.255s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.output: torch.Size([1, 79, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.297s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 79, 128]) + layer.0.v_cache: torch.Size([1, 4, 79, 128]) + layer.1.k_cache: torch.Size([1, 4, 79, 128]) + layer.1.v_cache: torch.Size([1, 4, 79, 128]) + layer.2.k_cache: torch.Size([1, 4, 79, 128]) + layer.2.v_cache: torch.Size([1, 4, 79, 128]) + layer.3.k_cache: torch.Size([1, 4, 79, 128]) + layer.3.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.k_cache: torch.Size([1, 4, 79, 128]) + layer.4.v_cache: torch.Size([1, 4, 79, 128]) + layer.4.output: torch.Size([1, 79, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14563359 29.38557531 + layer.0.v_cache 0.00001368 0.00431129 + layer.1.k_cache 0.05947259 3.89316066 + layer.1.v_cache 0.00000496 0.00156249 + layer.2.k_cache 0.00553216 0.47846232 + layer.2.v_cache 0.00001557 0.00456057 + layer.3.k_cache 0.04377548 2.00685139 + layer.3.v_cache 0.00001730 0.00545460 + layer.4.k_cache 0.00079035 0.10656152 + layer.4.v_cache 0.00005153 0.01074081 + layer.4.output 0.17347779 670.40065552 + ------------------------------------------------------------------------------------- + TOTAL 0.08645010 278.15893115 + (elements=687,616) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 687616 +Total Bytes 148488 +BPFP 1.7276 bits/point +EBPFP 3.4551 equivalent bits/point +MSE 278.158931 +---------------------- -------------------------------------------------------- +Time: 0.555s Load: 0.003s, Pack+Encode: 0.255s, Decode+Unpack: 0.297s +---------------------- -------------------------------------------------------- +💾 Converting with 278.1589 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-386.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-386.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-389.zst (57/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-389.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 66, 128) +Output shape: (1, 66, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.output: torch.Size([1, 66, 3584]) -> torch.Size([1, 1, 66, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,264B, BPFP=1.0095 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 13,928B, BPFP=3.2973 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 7,576B, BPFP=1.7936 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 12,756B, BPFP=3.0199 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,692B, BPFP=2.0578 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,364B, BPFP=2.9271 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 7,756B, BPFP=1.8362 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 12,868B, BPFP=3.0464 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,360B, BPFP=2.6894 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 12,700B, BPFP=3.0066 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,656B, BPFP=0.9692 +⌛️ [2/4] FRONTEND: Frontend time: 0.235s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.311s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12373332 32.30489095 + layer.0.v_cache 0.00001436 0.00507507 + layer.1.k_cache 0.01270404 3.53038441 + layer.1.v_cache 0.00000490 0.00167730 + layer.2.k_cache 0.00252672 0.47597166 + layer.2.v_cache 0.00001629 0.00513846 + layer.3.k_cache 0.05185594 2.42202481 + layer.3.v_cache 0.00001849 0.00596815 + layer.4.k_cache 0.00068458 0.11408612 + layer.4.v_cache 0.00004975 0.01128565 + layer.4.output 0.22283003 811.77313312 + ------------------------------------------------------------------------------------- + TOTAL 0.10302462 336.54637849 + (elements=574,464) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 574464 +Total Bytes 132920 +BPFP 1.8510 bits/point +EBPFP 3.7021 equivalent bits/point +MSE 336.546378 +---------------------- -------------------------------------------------------- +Time: 0.549s Load: 0.003s, Pack+Encode: 0.235s, Decode+Unpack: 0.311s +---------------------- -------------------------------------------------------- +💾 Converting with 336.5464 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-389.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-389.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-390.zst (58/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-390.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 84, 128) +Output shape: (1, 84, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) -> torch.Size([1, 1, 84, 512]) + layer.4.output: torch.Size([1, 84, 3584]) -> torch.Size([1, 1, 84, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,216B, BPFP=0.9702 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,192B, BPFP=3.0119 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,932B, BPFP=1.6615 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,640B, BPFP=2.9092 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,184B, BPFP=1.8943 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,180B, BPFP=2.8237 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,688B, BPFP=1.8021 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,716B, BPFP=2.9234 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,352B, BPFP=2.4836 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,216B, BPFP=2.8304 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 36,488B, BPFP=0.9696 +⌛️ [2/4] FRONTEND: Frontend time: 0.257s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.290s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 84, 128]) + layer.0.v_cache: torch.Size([1, 4, 84, 128]) + layer.1.k_cache: torch.Size([1, 4, 84, 128]) + layer.1.v_cache: torch.Size([1, 4, 84, 128]) + layer.2.k_cache: torch.Size([1, 4, 84, 128]) + layer.2.v_cache: torch.Size([1, 4, 84, 128]) + layer.3.k_cache: torch.Size([1, 4, 84, 128]) + layer.3.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.k_cache: torch.Size([1, 4, 84, 128]) + layer.4.v_cache: torch.Size([1, 4, 84, 128]) + layer.4.output: torch.Size([1, 84, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12002775 29.16767229 + layer.0.v_cache 0.00001466 0.00500134 + layer.1.k_cache 0.05271155 3.35052091 + layer.1.v_cache 0.00000524 0.00184866 + layer.2.k_cache 0.00808975 0.45381996 + layer.2.v_cache 0.00001826 0.00488381 + layer.3.k_cache 0.07501186 2.17286319 + layer.3.v_cache 0.00001895 0.00599277 + layer.4.k_cache 0.00071216 0.11685299 + layer.4.v_cache 0.00005170 0.01227983 + layer.4.output 0.17053346 631.15226403 + ------------------------------------------------------------------------------------- + TOTAL 0.08531742 261.96221082 + (elements=731,136) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 731136 +Total Bytes 161804 +BPFP 1.7704 bits/point +EBPFP 3.5409 equivalent bits/point +MSE 261.962211 +---------------------- -------------------------------------------------------- +Time: 0.552s Load: 0.005s, Pack+Encode: 0.257s, Decode+Unpack: 0.290s +---------------------- -------------------------------------------------------- +💾 Converting with 261.9622 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-390.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-390.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-395.zst (59/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-395.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 75, 128) +Output shape: (1, 75, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.0.v_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.1.k_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.1.v_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.2.k_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.2.v_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.3.k_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.3.v_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.4.k_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.4.v_cache: torch.Size([1, 4, 75, 128]) -> torch.Size([1, 1, 75, 512]) + layer.4.output: torch.Size([1, 75, 3584]) -> torch.Size([1, 1, 75, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,928B, BPFP=1.0267 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,060B, BPFP=3.1375 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,356B, BPFP=1.7408 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,296B, BPFP=2.9783 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,656B, BPFP=2.0117 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,864B, BPFP=2.8883 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,748B, BPFP=1.8225 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,144B, BPFP=2.9467 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,368B, BPFP=2.5767 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,080B, BPFP=2.9333 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,612B, BPFP=0.9408 +⌛️ [2/4] FRONTEND: Frontend time: 0.235s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 75, 128]) + layer.0.v_cache: torch.Size([1, 4, 75, 128]) + layer.1.k_cache: torch.Size([1, 4, 75, 128]) + layer.1.v_cache: torch.Size([1, 4, 75, 128]) + layer.2.k_cache: torch.Size([1, 4, 75, 128]) + layer.2.v_cache: torch.Size([1, 4, 75, 128]) + layer.3.k_cache: torch.Size([1, 4, 75, 128]) + layer.3.v_cache: torch.Size([1, 4, 75, 128]) + layer.4.k_cache: torch.Size([1, 4, 75, 128]) + layer.4.v_cache: torch.Size([1, 4, 75, 128]) + layer.4.output: torch.Size([1, 75, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.294s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 75, 128]) + layer.0.v_cache: torch.Size([1, 4, 75, 128]) + layer.1.k_cache: torch.Size([1, 4, 75, 128]) + layer.1.v_cache: torch.Size([1, 4, 75, 128]) + layer.2.k_cache: torch.Size([1, 4, 75, 128]) + layer.2.v_cache: torch.Size([1, 4, 75, 128]) + layer.3.k_cache: torch.Size([1, 4, 75, 128]) + layer.3.v_cache: torch.Size([1, 4, 75, 128]) + layer.4.k_cache: torch.Size([1, 4, 75, 128]) + layer.4.v_cache: torch.Size([1, 4, 75, 128]) + layer.4.output: torch.Size([1, 75, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08149677 31.55136393 + layer.0.v_cache 0.00001391 0.00512651 + layer.1.k_cache 0.05841089 3.39947144 + layer.1.v_cache 0.00000541 0.00184514 + layer.2.k_cache 0.00252637 0.47475057 + layer.2.v_cache 0.00001693 0.00505050 + layer.3.k_cache 0.02876982 2.49705770 + layer.3.v_cache 0.00001932 0.00578911 + layer.4.k_cache 0.00075246 0.11957258 + layer.4.v_cache 0.00005041 0.01238063 + layer.4.output 0.19266937 709.29976190 + ------------------------------------------------------------------------------------- + TOTAL 0.08945576 294.30416126 + (elements=652,800) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 652800 +Total Bytes 147112 +BPFP 1.8028 bits/point +EBPFP 3.6057 equivalent bits/point +MSE 294.304161 +---------------------- -------------------------------------------------------- +Time: 0.533s Load: 0.004s, Pack+Encode: 0.235s, Decode+Unpack: 0.294s +---------------------- -------------------------------------------------------- +💾 Converting with 294.3042 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-395.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-395.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-396.zst (60/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-396.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 68, 128) +Output shape: (1, 68, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.output: torch.Size([1, 68, 3584]) -> torch.Size([1, 1, 68, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,596B, BPFP=1.0561 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,164B, BPFP=3.2546 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,064B, BPFP=1.8529 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 12,860B, BPFP=2.9550 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,916B, BPFP=2.0487 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,612B, BPFP=2.8980 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,048B, BPFP=1.8493 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,136B, BPFP=3.0184 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,708B, BPFP=2.6903 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,144B, BPFP=3.0202 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,292B, BPFP=0.9944 +⌛️ [2/4] FRONTEND: Frontend time: 0.230s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.280s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10404026 29.66336957 + layer.0.v_cache 0.00001395 0.00508462 + layer.1.k_cache 0.01458383 3.72355831 + layer.1.v_cache 0.00000499 0.00171553 + layer.2.k_cache 0.00236740 0.46532334 + layer.2.v_cache 0.00001641 0.00495569 + layer.3.k_cache 0.03645969 2.03769280 + layer.3.v_cache 0.00002420 0.00572628 + layer.4.k_cache 0.00069996 0.11644650 + layer.4.v_cache 0.00005119 0.01211495 + layer.4.output 0.20980707 771.85241597 + ------------------------------------------------------------------------------------- + TOTAL 0.09570067 319.94134702 + (elements=591,872) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 591872 +Total Bytes 137540 +BPFP 1.8591 bits/point +EBPFP 3.7181 equivalent bits/point +MSE 319.941347 +---------------------- -------------------------------------------------------- +Time: 0.513s Load: 0.003s, Pack+Encode: 0.230s, Decode+Unpack: 0.280s +---------------------- -------------------------------------------------------- +💾 Converting with 319.9413 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-396.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-396.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-401.zst (61/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-401.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 70, 128) +Output shape: (1, 70, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.output: torch.Size([1, 70, 3584]) -> torch.Size([1, 1, 70, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,672B, BPFP=1.0429 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,620B, BPFP=3.2634 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,180B, BPFP=1.8259 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,580B, BPFP=3.0312 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,076B, BPFP=2.0259 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,200B, BPFP=2.9464 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,560B, BPFP=1.9107 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,528B, BPFP=3.0196 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,888B, BPFP=2.6536 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,432B, BPFP=2.9982 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,204B, BPFP=1.0269 +⌛️ [2/4] FRONTEND: Frontend time: 0.243s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.324s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11725240 31.01826521 + layer.0.v_cache 0.00001567 0.00507255 + layer.1.k_cache 0.03693308 3.74914464 + layer.1.v_cache 0.00000533 0.00211390 + layer.2.k_cache 0.00240566 0.46050126 + layer.2.v_cache 0.00001850 0.00576908 + layer.3.k_cache 0.06751672 2.18178231 + layer.3.v_cache 0.00001909 0.00588159 + layer.4.k_cache 0.00067984 0.11870153 + layer.4.v_cache 0.00004901 0.01222210 + layer.4.output 0.20094273 769.53258929 + ------------------------------------------------------------------------------------- + TOTAL 0.09597026 319.07573995 + (elements=609,280) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 609280 +Total Bytes 142940 +BPFP 1.8768 bits/point +EBPFP 3.7537 equivalent bits/point +MSE 319.075740 +---------------------- -------------------------------------------------------- +Time: 0.571s Load: 0.003s, Pack+Encode: 0.243s, Decode+Unpack: 0.324s +---------------------- -------------------------------------------------------- +💾 Converting with 319.0757 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-401.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-401.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-408.zst (62/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-408.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 70, 128) +Output shape: (1, 70, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.output: torch.Size([1, 70, 3584]) -> torch.Size([1, 1, 70, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,640B, BPFP=1.0357 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,436B, BPFP=3.2223 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,020B, BPFP=1.7902 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,184B, BPFP=2.9429 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,868B, BPFP=1.9795 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,708B, BPFP=2.8366 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,216B, BPFP=1.8339 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,104B, BPFP=2.9250 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,616B, BPFP=2.5929 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 12,872B, BPFP=2.8732 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,196B, BPFP=0.9310 +⌛️ [2/4] FRONTEND: Frontend time: 0.237s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.302s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.07685234 30.04423828 + layer.0.v_cache 0.00001390 0.00511100 + layer.1.k_cache 0.01187498 3.55555899 + layer.1.v_cache 0.00000934 0.00193458 + layer.2.k_cache 0.00239603 0.48761684 + layer.2.v_cache 0.00001621 0.00464546 + layer.3.k_cache 0.07241479 2.05029885 + layer.3.v_cache 0.00001716 0.00602002 + layer.4.k_cache 0.00072067 0.11669158 + layer.4.v_cache 0.00004958 0.01125308 + layer.4.output 0.21417375 768.62704082 + ------------------------------------------------------------------------------------- + TOTAL 0.09785772 318.62780320 + (elements=609,280) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 609280 +Total Bytes 136860 +BPFP 1.7970 bits/point +EBPFP 3.5940 equivalent bits/point +MSE 318.627803 +---------------------- -------------------------------------------------------- +Time: 0.542s Load: 0.003s, Pack+Encode: 0.237s, Decode+Unpack: 0.302s +---------------------- -------------------------------------------------------- +💾 Converting with 318.6278 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-408.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-408.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-417.zst (63/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-417.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 76, 128) +Output shape: (1, 76, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.output: torch.Size([1, 76, 3584]) -> torch.Size([1, 1, 76, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,032B, BPFP=1.0345 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,340B, BPFP=3.1538 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,508B, BPFP=1.7492 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,684B, BPFP=3.0189 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,700B, BPFP=1.9942 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,004B, BPFP=2.8791 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,896B, BPFP=1.8289 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,648B, BPFP=3.0115 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,628B, BPFP=2.5962 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,388B, BPFP=2.9581 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,724B, BPFP=0.9905 +⌛️ [2/4] FRONTEND: Frontend time: 0.264s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.output: torch.Size([1, 76, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.317s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.output: torch.Size([1, 76, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09264422 30.52364309 + layer.0.v_cache 0.00001680 0.00552433 + layer.1.k_cache 0.05606840 3.54621526 + layer.1.v_cache 0.00000542 0.00220968 + layer.2.k_cache 0.00241068 0.48882876 + layer.2.v_cache 0.00001702 0.00523206 + layer.3.k_cache 0.15270803 2.47155300 + layer.3.v_cache 0.00001849 0.00704421 + layer.4.k_cache 0.00070970 0.13093751 + layer.4.v_cache 0.00005539 0.01330728 + layer.4.output 0.18838861 684.86894972 + ------------------------------------------------------------------------------------- + TOTAL 0.09549261 284.19277313 + (elements=661,504) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 661504 +Total Bytes 151552 +BPFP 1.8328 bits/point +EBPFP 3.6656 equivalent bits/point +MSE 284.192773 +---------------------- -------------------------------------------------------- +Time: 0.586s Load: 0.005s, Pack+Encode: 0.264s, Decode+Unpack: 0.317s +---------------------- -------------------------------------------------------- +💾 Converting with 284.1928 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-417.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-417.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-422.zst (64/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-422.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,016B, BPFP=1.0179 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,216B, BPFP=3.0877 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,516B, BPFP=1.7281 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,460B, BPFP=2.9343 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,740B, BPFP=1.9765 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,160B, BPFP=2.8734 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,896B, BPFP=1.8052 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,728B, BPFP=2.9886 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,736B, BPFP=2.5844 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,368B, BPFP=2.9156 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,120B, BPFP=0.8442 +⌛️ [2/4] FRONTEND: Frontend time: 0.259s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.326s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10144231 32.16756291 + layer.0.v_cache 0.00001330 0.00484800 + layer.1.k_cache 0.09893422 3.67161065 + layer.1.v_cache 0.00000513 0.00182994 + layer.2.k_cache 0.00368564 0.52008889 + layer.2.v_cache 0.00001777 0.00500579 + layer.3.k_cache 0.06341467 2.65881189 + layer.3.v_cache 0.00001739 0.00592533 + layer.4.k_cache 0.00073693 0.12447991 + layer.4.v_cache 0.00004994 0.01136727 + layer.4.output 0.18261438 698.04904917 + ------------------------------------------------------------------------------------- + TOTAL 0.09097753 289.73616910 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 146956 +BPFP 1.7542 bits/point +EBPFP 3.5083 equivalent bits/point +MSE 289.736169 +---------------------- -------------------------------------------------------- +Time: 0.589s Load: 0.003s, Pack+Encode: 0.259s, Decode+Unpack: 0.326s +---------------------- -------------------------------------------------------- +💾 Converting with 289.7362 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-422.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-422.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-432.zst (65/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-432.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 87, 128) +Output shape: (1, 87, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.0.v_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.1.k_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.1.v_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.2.k_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.2.v_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.3.k_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.3.v_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.4.k_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.4.v_cache: torch.Size([1, 4, 87, 128]) -> torch.Size([1, 1, 87, 512]) + layer.4.output: torch.Size([1, 87, 3584]) -> torch.Size([1, 1, 87, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,228B, BPFP=0.9389 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,500B, BPFP=2.7838 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,868B, BPFP=1.5927 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,808B, BPFP=2.6595 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,744B, BPFP=1.7500 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,396B, BPFP=2.5855 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,172B, BPFP=1.6473 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,076B, BPFP=2.7076 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,236B, BPFP=2.3772 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,600B, BPFP=2.6221 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,052B, BPFP=0.7967 +⌛️ [2/4] FRONTEND: Frontend time: 0.268s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 87, 128]) + layer.0.v_cache: torch.Size([1, 4, 87, 128]) + layer.1.k_cache: torch.Size([1, 4, 87, 128]) + layer.1.v_cache: torch.Size([1, 4, 87, 128]) + layer.2.k_cache: torch.Size([1, 4, 87, 128]) + layer.2.v_cache: torch.Size([1, 4, 87, 128]) + layer.3.k_cache: torch.Size([1, 4, 87, 128]) + layer.3.v_cache: torch.Size([1, 4, 87, 128]) + layer.4.k_cache: torch.Size([1, 4, 87, 128]) + layer.4.v_cache: torch.Size([1, 4, 87, 128]) + layer.4.output: torch.Size([1, 87, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.328s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 87, 128]) + layer.0.v_cache: torch.Size([1, 4, 87, 128]) + layer.1.k_cache: torch.Size([1, 4, 87, 128]) + layer.1.v_cache: torch.Size([1, 4, 87, 128]) + layer.2.k_cache: torch.Size([1, 4, 87, 128]) + layer.2.v_cache: torch.Size([1, 4, 87, 128]) + layer.3.k_cache: torch.Size([1, 4, 87, 128]) + layer.3.v_cache: torch.Size([1, 4, 87, 128]) + layer.4.k_cache: torch.Size([1, 4, 87, 128]) + layer.4.v_cache: torch.Size([1, 4, 87, 128]) + layer.4.output: torch.Size([1, 87, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08947531 27.54075184 + layer.0.v_cache 0.00001432 0.00437592 + layer.1.k_cache 0.05320675 3.48259937 + layer.1.v_cache 0.00000497 0.00164942 + layer.2.k_cache 0.00562141 0.47605142 + layer.2.v_cache 0.00001574 0.00396611 + layer.3.k_cache 0.02742601 2.12240776 + layer.3.v_cache 0.00001759 0.00494611 + layer.4.k_cache 0.00114861 0.11444831 + layer.4.v_cache 0.00004734 0.01012181 + layer.4.output 0.15628038 612.85688629 + ------------------------------------------------------------------------------------- + TOTAL 0.07476122 254.33879542 + (elements=757,248) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 757248 +Total Bytes 151680 +BPFP 1.6024 bits/point +EBPFP 3.2049 equivalent bits/point +MSE 254.338795 +---------------------- -------------------------------------------------------- +Time: 0.601s Load: 0.005s, Pack+Encode: 0.268s, Decode+Unpack: 0.328s +---------------------- -------------------------------------------------------- +💾 Converting with 254.3388 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-432.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-432.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-435.zst (66/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-435.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 92, 128) +Output shape: (1, 92, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.output: torch.Size([1, 92, 3584]) -> torch.Size([1, 1, 92, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,600B, BPFP=0.9511 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,304B, BPFP=2.7690 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,596B, BPFP=1.6298 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,684B, BPFP=2.6637 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,412B, BPFP=1.7683 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,312B, BPFP=2.6005 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,888B, BPFP=1.6793 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,748B, BPFP=2.6746 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,768B, BPFP=2.3383 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,100B, BPFP=2.5645 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 36,252B, BPFP=0.8796 +⌛️ [2/4] FRONTEND: Frontend time: 0.235s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.280s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12311893 28.71913479 + layer.0.v_cache 0.00001357 0.00450688 + layer.1.k_cache 0.03743907 3.06844827 + layer.1.v_cache 0.00000515 0.00174015 + layer.2.k_cache 0.00511221 0.47029383 + layer.2.v_cache 0.00001545 0.00437802 + layer.3.k_cache 0.04225797 2.33412635 + layer.3.v_cache 0.00001799 0.00539186 + layer.4.k_cache 0.00119426 0.10964052 + layer.4.v_cache 0.00004621 0.01019545 + layer.4.output 0.14783020 573.96719720 + ------------------------------------------------------------------------------------- + TOTAL 0.07317837 238.38224921 + (elements=800,768) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 800768 +Total Bytes 163664 +BPFP 1.6351 bits/point +EBPFP 3.2701 equivalent bits/point +MSE 238.382249 +---------------------- -------------------------------------------------------- +Time: 0.521s Load: 0.005s, Pack+Encode: 0.235s, Decode+Unpack: 0.280s +---------------------- -------------------------------------------------------- +💾 Converting with 238.3822 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-435.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-435.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-438.zst (67/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-438.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 102, 128) +Output shape: (1, 102, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) -> torch.Size([1, 1, 102, 512]) + layer.4.output: torch.Size([1, 102, 3584]) -> torch.Size([1, 1, 102, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,804B, BPFP=0.8891 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 18,408B, BPFP=2.8199 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,092B, BPFP=1.5460 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 17,564B, BPFP=2.6906 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,136B, BPFP=1.7059 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 17,188B, BPFP=2.6330 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,632B, BPFP=1.6287 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 17,728B, BPFP=2.7157 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 15,072B, BPFP=2.3088 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 17,048B, BPFP=2.6115 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 40,340B, BPFP=0.8828 +⌛️ [2/4] FRONTEND: Frontend time: 0.243s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.output: torch.Size([1, 102, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.317s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 102, 128]) + layer.0.v_cache: torch.Size([1, 4, 102, 128]) + layer.1.k_cache: torch.Size([1, 4, 102, 128]) + layer.1.v_cache: torch.Size([1, 4, 102, 128]) + layer.2.k_cache: torch.Size([1, 4, 102, 128]) + layer.2.v_cache: torch.Size([1, 4, 102, 128]) + layer.3.k_cache: torch.Size([1, 4, 102, 128]) + layer.3.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.k_cache: torch.Size([1, 4, 102, 128]) + layer.4.v_cache: torch.Size([1, 4, 102, 128]) + layer.4.output: torch.Size([1, 102, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09293266 27.15399050 + layer.0.v_cache 0.00001503 0.00489529 + layer.1.k_cache 0.07854857 3.42735380 + layer.1.v_cache 0.00000559 0.00170304 + layer.2.k_cache 0.00650115 0.44379343 + layer.2.v_cache 0.00001763 0.00459020 + layer.3.k_cache 0.03715762 2.11948679 + layer.3.v_cache 0.00001788 0.00545397 + layer.4.k_cache 0.00073365 0.10251079 + layer.4.v_cache 0.00004630 0.01030101 + layer.4.output 11.17680568 522.51221113 + ------------------------------------------------------------------------------------- + TOTAL 4.61491858 217.10938569 + (elements=887,808) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 887808 +Total Bytes 181012 +BPFP 1.6311 bits/point +EBPFP 3.2622 equivalent bits/point +MSE 217.109386 +---------------------- -------------------------------------------------------- +Time: 0.565s Load: 0.004s, Pack+Encode: 0.243s, Decode+Unpack: 0.317s +---------------------- -------------------------------------------------------- +💾 Converting with 217.1094 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-438.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-438.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-44.zst (68/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-44.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 96, 128) +Output shape: (1, 96, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.0.v_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.1.k_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.1.v_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.2.k_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.2.v_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.3.k_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.3.v_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.4.k_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.4.v_cache: torch.Size([1, 4, 96, 128]) -> torch.Size([1, 1, 96, 512]) + layer.4.output: torch.Size([1, 96, 3584]) -> torch.Size([1, 1, 96, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,280B, BPFP=0.8594 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,512B, BPFP=2.8503 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,728B, BPFP=1.5833 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,844B, BPFP=2.7415 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,792B, BPFP=1.7565 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,476B, BPFP=2.6816 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,304B, BPFP=1.6771 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,896B, BPFP=2.7500 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,668B, BPFP=2.3874 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,320B, BPFP=2.6562 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 40,528B, BPFP=0.9423 +⌛️ [2/4] FRONTEND: Frontend time: 0.247s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 96, 128]) + layer.0.v_cache: torch.Size([1, 4, 96, 128]) + layer.1.k_cache: torch.Size([1, 4, 96, 128]) + layer.1.v_cache: torch.Size([1, 4, 96, 128]) + layer.2.k_cache: torch.Size([1, 4, 96, 128]) + layer.2.v_cache: torch.Size([1, 4, 96, 128]) + layer.3.k_cache: torch.Size([1, 4, 96, 128]) + layer.3.v_cache: torch.Size([1, 4, 96, 128]) + layer.4.k_cache: torch.Size([1, 4, 96, 128]) + layer.4.v_cache: torch.Size([1, 4, 96, 128]) + layer.4.output: torch.Size([1, 96, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.257s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 96, 128]) + layer.0.v_cache: torch.Size([1, 4, 96, 128]) + layer.1.k_cache: torch.Size([1, 4, 96, 128]) + layer.1.v_cache: torch.Size([1, 4, 96, 128]) + layer.2.k_cache: torch.Size([1, 4, 96, 128]) + layer.2.v_cache: torch.Size([1, 4, 96, 128]) + layer.3.k_cache: torch.Size([1, 4, 96, 128]) + layer.3.v_cache: torch.Size([1, 4, 96, 128]) + layer.4.k_cache: torch.Size([1, 4, 96, 128]) + layer.4.v_cache: torch.Size([1, 4, 96, 128]) + layer.4.output: torch.Size([1, 96, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10158994 27.57086182 + layer.0.v_cache 0.00001467 0.00545673 + layer.1.k_cache 0.06416908 3.23317655 + layer.1.v_cache 0.00000542 0.00191452 + layer.2.k_cache 0.00400108 0.47325699 + layer.2.v_cache 0.00001872 0.00539309 + layer.3.k_cache 0.01491791 2.34545898 + layer.3.v_cache 0.00001877 0.00628412 + layer.4.k_cache 0.00070988 0.12258190 + layer.4.v_cache 0.00004832 0.01192896 + layer.4.output 0.14169503 538.81542969 + ------------------------------------------------------------------------------------- + TOTAL 0.06925641 223.85201891 + (elements=835,584) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 835584 +Total Bytes 175348 +BPFP 1.6788 bits/point +EBPFP 3.3576 equivalent bits/point +MSE 223.852019 +---------------------- -------------------------------------------------------- +Time: 0.509s Load: 0.005s, Pack+Encode: 0.247s, Decode+Unpack: 0.257s +---------------------- -------------------------------------------------------- +💾 Converting with 223.8520 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-44.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-44.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-440.zst (69/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-440.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 72, 128) +Output shape: (1, 72, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.output: torch.Size([1, 72, 3584]) -> torch.Size([1, 1, 72, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,816B, BPFP=1.0451 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,888B, BPFP=3.2309 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,168B, BPFP=1.7726 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,876B, BPFP=3.0113 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,396B, BPFP=2.0391 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,476B, BPFP=2.9245 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,464B, BPFP=1.8368 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,816B, BPFP=2.9983 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,228B, BPFP=2.6536 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,800B, BPFP=2.9948 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,132B, BPFP=0.9342 +⌛️ [2/4] FRONTEND: Frontend time: 0.238s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.output: torch.Size([1, 72, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.305s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.output: torch.Size([1, 72, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10321087 30.99604628 + layer.0.v_cache 0.00001557 0.00522426 + layer.1.k_cache 0.06175087 3.98041068 + layer.1.v_cache 0.00000571 0.00192902 + layer.2.k_cache 0.01329466 0.51556487 + layer.2.v_cache 0.00001784 0.00531070 + layer.3.k_cache 0.05039416 2.49405628 + layer.3.v_cache 0.00001745 0.00597610 + layer.4.k_cache 0.00079786 0.12055018 + layer.4.v_cache 0.00005071 0.01211049 + layer.4.output 0.18870974 741.48015873 + ------------------------------------------------------------------------------------- + TOTAL 0.09120729 307.55872294 + (elements=626,688) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 626688 +Total Bytes 143060 +BPFP 1.8262 bits/point +EBPFP 3.6525 equivalent bits/point +MSE 307.558723 +---------------------- -------------------------------------------------------- +Time: 0.546s Load: 0.003s, Pack+Encode: 0.238s, Decode+Unpack: 0.305s +---------------------- -------------------------------------------------------- +💾 Converting with 307.5587 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-440.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-440.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-443.zst (70/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-443.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 72, 128) +Output shape: (1, 72, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) -> torch.Size([1, 1, 72, 512]) + layer.4.output: torch.Size([1, 72, 3584]) -> torch.Size([1, 1, 72, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,868B, BPFP=1.0564 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,356B, BPFP=3.1155 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,116B, BPFP=1.7613 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,400B, BPFP=2.9080 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,896B, BPFP=1.9306 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,740B, BPFP=2.7648 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,328B, BPFP=1.8073 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,180B, BPFP=2.8602 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,784B, BPFP=2.5573 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,228B, BPFP=2.8707 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,664B, BPFP=0.9506 +⌛️ [2/4] FRONTEND: Frontend time: 0.244s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.output: torch.Size([1, 72, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.311s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 72, 128]) + layer.0.v_cache: torch.Size([1, 4, 72, 128]) + layer.1.k_cache: torch.Size([1, 4, 72, 128]) + layer.1.v_cache: torch.Size([1, 4, 72, 128]) + layer.2.k_cache: torch.Size([1, 4, 72, 128]) + layer.2.v_cache: torch.Size([1, 4, 72, 128]) + layer.3.k_cache: torch.Size([1, 4, 72, 128]) + layer.3.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.k_cache: torch.Size([1, 4, 72, 128]) + layer.4.v_cache: torch.Size([1, 4, 72, 128]) + layer.4.output: torch.Size([1, 72, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09666174 30.32084147 + layer.0.v_cache 0.00001715 0.00478718 + layer.1.k_cache 0.08476152 3.76403130 + layer.1.v_cache 0.00000506 0.00171133 + layer.2.k_cache 0.00238475 0.51807462 + layer.2.v_cache 0.00001567 0.00465663 + layer.3.k_cache 0.07332428 2.42850494 + layer.3.v_cache 0.00001796 0.00552482 + layer.4.k_cache 0.00072290 0.11324281 + layer.4.v_cache 0.00005092 0.01066830 + layer.4.output 0.18875189 742.04024058 + ------------------------------------------------------------------------------------- + TOTAL 0.09289560 307.73257220 + (elements=626,688) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 626688 +Total Bytes 139560 +BPFP 1.7816 bits/point +EBPFP 3.5631 equivalent bits/point +MSE 307.732572 +---------------------- -------------------------------------------------------- +Time: 0.559s Load: 0.003s, Pack+Encode: 0.244s, Decode+Unpack: 0.311s +---------------------- -------------------------------------------------------- +💾 Converting with 307.7326 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-443.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-443.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-444.zst (71/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-444.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 68, 128) +Output shape: (1, 68, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) -> torch.Size([1, 1, 68, 512]) + layer.4.output: torch.Size([1, 68, 3584]) -> torch.Size([1, 1, 68, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,572B, BPFP=1.0506 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,320B, BPFP=3.2904 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,008B, BPFP=1.8401 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,212B, BPFP=3.0358 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,120B, BPFP=2.0956 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,708B, BPFP=2.9200 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,244B, BPFP=1.8943 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 12,944B, BPFP=2.9743 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,820B, BPFP=2.7160 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,344B, BPFP=3.0662 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 31,720B, BPFP=1.0412 +⌛️ [2/4] FRONTEND: Frontend time: 0.231s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.294s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 68, 128]) + layer.0.v_cache: torch.Size([1, 4, 68, 128]) + layer.1.k_cache: torch.Size([1, 4, 68, 128]) + layer.1.v_cache: torch.Size([1, 4, 68, 128]) + layer.2.k_cache: torch.Size([1, 4, 68, 128]) + layer.2.v_cache: torch.Size([1, 4, 68, 128]) + layer.3.k_cache: torch.Size([1, 4, 68, 128]) + layer.3.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.k_cache: torch.Size([1, 4, 68, 128]) + layer.4.v_cache: torch.Size([1, 4, 68, 128]) + layer.4.output: torch.Size([1, 68, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10601801 29.70063153 + layer.0.v_cache 0.00001817 0.00507606 + layer.1.k_cache 0.01696859 3.36008229 + layer.1.v_cache 0.00000501 0.00188885 + layer.2.k_cache 0.00274148 0.49553052 + layer.2.v_cache 0.00001757 0.00524653 + layer.3.k_cache 0.01706722 2.22805248 + layer.3.v_cache 0.00001843 0.00589928 + layer.4.k_cache 0.00080906 0.11801397 + layer.4.v_cache 0.00005076 0.01162994 + layer.4.output 0.19983917 776.96139706 + ------------------------------------------------------------------------------------- + TOTAL 0.09074050 322.03893123 + (elements=591,872) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 591872 +Total Bytes 140012 +BPFP 1.8925 bits/point +EBPFP 3.7849 equivalent bits/point +MSE 322.038931 +---------------------- -------------------------------------------------------- +Time: 0.528s Load: 0.003s, Pack+Encode: 0.231s, Decode+Unpack: 0.294s +---------------------- -------------------------------------------------------- +💾 Converting with 322.0389 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-444.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-444.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-452.zst (72/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-452.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 65, 128) +Output shape: (1, 65, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.0.v_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.1.k_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.1.v_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.2.k_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.2.v_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.3.k_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.3.v_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.4.k_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.4.v_cache: torch.Size([1, 4, 65, 128]) -> torch.Size([1, 1, 65, 512]) + layer.4.output: torch.Size([1, 65, 3584]) -> torch.Size([1, 1, 65, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,144B, BPFP=0.9962 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 13,256B, BPFP=3.1865 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 7,132B, BPFP=1.7144 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 12,220B, BPFP=2.9375 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,388B, BPFP=2.0163 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 11,996B, BPFP=2.8837 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 7,416B, BPFP=1.7827 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 12,440B, BPFP=2.9904 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 10,908B, BPFP=2.6221 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 12,892B, BPFP=3.0990 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,172B, BPFP=0.9674 +⌛️ [2/4] FRONTEND: Frontend time: 0.247s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 65, 128]) + layer.0.v_cache: torch.Size([1, 4, 65, 128]) + layer.1.k_cache: torch.Size([1, 4, 65, 128]) + layer.1.v_cache: torch.Size([1, 4, 65, 128]) + layer.2.k_cache: torch.Size([1, 4, 65, 128]) + layer.2.v_cache: torch.Size([1, 4, 65, 128]) + layer.3.k_cache: torch.Size([1, 4, 65, 128]) + layer.3.v_cache: torch.Size([1, 4, 65, 128]) + layer.4.k_cache: torch.Size([1, 4, 65, 128]) + layer.4.v_cache: torch.Size([1, 4, 65, 128]) + layer.4.output: torch.Size([1, 65, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.325s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 65, 128]) + layer.0.v_cache: torch.Size([1, 4, 65, 128]) + layer.1.k_cache: torch.Size([1, 4, 65, 128]) + layer.1.v_cache: torch.Size([1, 4, 65, 128]) + layer.2.k_cache: torch.Size([1, 4, 65, 128]) + layer.2.v_cache: torch.Size([1, 4, 65, 128]) + layer.3.k_cache: torch.Size([1, 4, 65, 128]) + layer.3.v_cache: torch.Size([1, 4, 65, 128]) + layer.4.k_cache: torch.Size([1, 4, 65, 128]) + layer.4.v_cache: torch.Size([1, 4, 65, 128]) + layer.4.output: torch.Size([1, 65, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09579474 32.50241136 + layer.0.v_cache 0.00001312 0.00507681 + layer.1.k_cache 0.01648773 3.71768940 + layer.1.v_cache 0.00000712 0.00181446 + layer.2.k_cache 0.00431219 0.44795638 + layer.2.v_cache 0.00001803 0.00544173 + layer.3.k_cache 0.03460605 2.45576735 + layer.3.v_cache 0.00001721 0.00692884 + layer.4.k_cache 0.00068162 0.12035987 + layer.4.v_cache 0.00004727 0.01247592 + layer.4.output 0.20898957 822.45364011 + ------------------------------------------------------------------------------------- + TOTAL 0.09499483 340.96772958 + (elements=565,760) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 565760 +Total Bytes 128964 +BPFP 1.8236 bits/point +EBPFP 3.6472 equivalent bits/point +MSE 340.967730 +---------------------- -------------------------------------------------------- +Time: 0.575s Load: 0.003s, Pack+Encode: 0.247s, Decode+Unpack: 0.325s +---------------------- -------------------------------------------------------- +💾 Converting with 340.9677 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-452.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-452.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-462.zst (73/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-462.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 41, 128) +Output shape: (1, 41, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.0.v_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.1.k_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.1.v_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.2.k_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.2.v_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.3.k_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.3.v_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.4.k_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.4.v_cache: torch.Size([1, 4, 41, 128]) -> torch.Size([1, 1, 41, 512]) + layer.4.output: torch.Size([1, 41, 3584]) -> torch.Size([1, 1, 41, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,196B, BPFP=1.2180 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 8,400B, BPFP=3.2012 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,012B, BPFP=1.9101 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 8,144B, BPFP=3.1037 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 5,512B, BPFP=2.1006 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 7,968B, BPFP=3.0366 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,136B, BPFP=1.9573 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 8,088B, BPFP=3.0823 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 7,096B, BPFP=2.7043 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 7,996B, BPFP=3.0473 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 22,396B, BPFP=1.2193 +⌛️ [2/4] FRONTEND: Frontend time: 0.170s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 41, 128]) + layer.0.v_cache: torch.Size([1, 4, 41, 128]) + layer.1.k_cache: torch.Size([1, 4, 41, 128]) + layer.1.v_cache: torch.Size([1, 4, 41, 128]) + layer.2.k_cache: torch.Size([1, 4, 41, 128]) + layer.2.v_cache: torch.Size([1, 4, 41, 128]) + layer.3.k_cache: torch.Size([1, 4, 41, 128]) + layer.3.v_cache: torch.Size([1, 4, 41, 128]) + layer.4.k_cache: torch.Size([1, 4, 41, 128]) + layer.4.v_cache: torch.Size([1, 4, 41, 128]) + layer.4.output: torch.Size([1, 41, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.174s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 41, 128]) + layer.0.v_cache: torch.Size([1, 4, 41, 128]) + layer.1.k_cache: torch.Size([1, 4, 41, 128]) + layer.1.v_cache: torch.Size([1, 4, 41, 128]) + layer.2.k_cache: torch.Size([1, 4, 41, 128]) + layer.2.v_cache: torch.Size([1, 4, 41, 128]) + layer.3.k_cache: torch.Size([1, 4, 41, 128]) + layer.3.v_cache: torch.Size([1, 4, 41, 128]) + layer.4.k_cache: torch.Size([1, 4, 41, 128]) + layer.4.v_cache: torch.Size([1, 4, 41, 128]) + layer.4.output: torch.Size([1, 41, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10098721 38.86802115 + layer.0.v_cache 0.00001362 0.00671504 + layer.1.k_cache 0.01708067 4.25553410 + layer.1.v_cache 0.00000533 0.00230655 + layer.2.k_cache 0.00250472 0.64578442 + layer.2.v_cache 0.00001755 0.00716347 + layer.3.k_cache 0.08199174 2.97021987 + layer.3.v_cache 0.00001815 0.00751317 + layer.4.k_cache 0.00059437 0.14012644 + layer.4.v_cache 0.00005232 0.01630310 + layer.4.output 0.35770382 1304.07295296 + ------------------------------------------------------------------------------------- + TOTAL 0.15924661 539.73119753 + (elements=356,864) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 356864 +Total Bytes 88944 +BPFP 1.9939 bits/point +EBPFP 3.9878 equivalent bits/point +MSE 539.731198 +---------------------- -------------------------------------------------------- +Time: 0.347s Load: 0.003s, Pack+Encode: 0.170s, Decode+Unpack: 0.174s +---------------------- -------------------------------------------------------- +💾 Converting with 539.7312 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-462.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-462.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-464.zst (74/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-464.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 52, 128) +Output shape: (1, 52, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) -> torch.Size([1, 1, 52, 512]) + layer.4.output: torch.Size([1, 52, 3584]) -> torch.Size([1, 1, 52, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,596B, BPFP=1.0805 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,036B, BPFP=3.0156 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,760B, BPFP=1.7308 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,716B, BPFP=2.9195 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,524B, BPFP=1.9603 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,508B, BPFP=2.8570 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,044B, BPFP=1.8161 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,664B, BPFP=2.9038 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,376B, BPFP=2.5168 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,320B, BPFP=2.8005 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,088B, BPFP=1.1198 +⌛️ [2/4] FRONTEND: Frontend time: 0.156s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.168s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 52, 128]) + layer.0.v_cache: torch.Size([1, 4, 52, 128]) + layer.1.k_cache: torch.Size([1, 4, 52, 128]) + layer.1.v_cache: torch.Size([1, 4, 52, 128]) + layer.2.k_cache: torch.Size([1, 4, 52, 128]) + layer.2.v_cache: torch.Size([1, 4, 52, 128]) + layer.3.k_cache: torch.Size([1, 4, 52, 128]) + layer.3.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.k_cache: torch.Size([1, 4, 52, 128]) + layer.4.v_cache: torch.Size([1, 4, 52, 128]) + layer.4.output: torch.Size([1, 52, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09210638 30.63103544 + layer.0.v_cache 0.00001369 0.00555150 + layer.1.k_cache 0.01682195 3.81465736 + layer.1.v_cache 0.00000520 0.00223944 + layer.2.k_cache 0.00487203 0.47149112 + layer.2.v_cache 0.00001755 0.00624023 + layer.3.k_cache 0.04965206 2.24995789 + layer.3.v_cache 0.00001944 0.00685037 + layer.4.k_cache 0.00060442 0.13295715 + layer.4.v_cache 0.00005207 0.01357653 + layer.4.output 0.26182200 996.53725962 + ------------------------------------------------------------------------------------- + TOTAL 0.11746581 412.53502202 + (elements=452,608) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 452608 +Total Bytes 104632 +BPFP 1.8494 bits/point +EBPFP 3.6988 equivalent bits/point +MSE 412.535022 +---------------------- -------------------------------------------------------- +Time: 0.327s Load: 0.003s, Pack+Encode: 0.156s, Decode+Unpack: 0.168s +---------------------- -------------------------------------------------------- +💾 Converting with 412.5350 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-464.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-464.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-467.zst (75/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-467.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 47, 128) +Output shape: (1, 47, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.0.v_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.1.k_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.1.v_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.2.k_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.2.v_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.3.k_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.3.v_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.4.k_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.4.v_cache: torch.Size([1, 4, 47, 128]) -> torch.Size([1, 1, 47, 512]) + layer.4.output: torch.Size([1, 47, 3584]) -> torch.Size([1, 1, 47, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,308B, BPFP=1.0997 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,584B, BPFP=3.1862 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,532B, BPFP=1.8391 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,288B, BPFP=3.0878 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,124B, BPFP=2.0359 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,060B, BPFP=3.0120 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,780B, BPFP=1.9215 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,060B, BPFP=3.0120 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,116B, BPFP=2.6981 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 8,980B, BPFP=2.9854 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 24,760B, BPFP=1.1759 +⌛️ [2/4] FRONTEND: Frontend time: 0.166s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 47, 128]) + layer.0.v_cache: torch.Size([1, 4, 47, 128]) + layer.1.k_cache: torch.Size([1, 4, 47, 128]) + layer.1.v_cache: torch.Size([1, 4, 47, 128]) + layer.2.k_cache: torch.Size([1, 4, 47, 128]) + layer.2.v_cache: torch.Size([1, 4, 47, 128]) + layer.3.k_cache: torch.Size([1, 4, 47, 128]) + layer.3.v_cache: torch.Size([1, 4, 47, 128]) + layer.4.k_cache: torch.Size([1, 4, 47, 128]) + layer.4.v_cache: torch.Size([1, 4, 47, 128]) + layer.4.output: torch.Size([1, 47, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.173s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 47, 128]) + layer.0.v_cache: torch.Size([1, 4, 47, 128]) + layer.1.k_cache: torch.Size([1, 4, 47, 128]) + layer.1.v_cache: torch.Size([1, 4, 47, 128]) + layer.2.k_cache: torch.Size([1, 4, 47, 128]) + layer.2.v_cache: torch.Size([1, 4, 47, 128]) + layer.3.k_cache: torch.Size([1, 4, 47, 128]) + layer.3.v_cache: torch.Size([1, 4, 47, 128]) + layer.4.k_cache: torch.Size([1, 4, 47, 128]) + layer.4.v_cache: torch.Size([1, 4, 47, 128]) + layer.4.output: torch.Size([1, 47, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08461995 34.51547176 + layer.0.v_cache 0.00001366 0.00574888 + layer.1.k_cache 0.01838745 3.95314708 + layer.1.v_cache 0.00000609 0.00254708 + layer.2.k_cache 0.00780689 0.54675058 + layer.2.v_cache 0.00001932 0.00636746 + layer.3.k_cache 0.04292001 2.23009442 + layer.3.v_cache 0.00001942 0.00714112 + layer.4.k_cache 0.00059704 0.12630253 + layer.4.v_cache 0.00004850 0.01437597 + layer.4.output 0.29441516 1100.89190729 + ------------------------------------------------------------------------------------- + TOTAL 0.13031438 455.74419400 + (elements=409,088) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 409088 +Total Bytes 99592 +BPFP 1.9476 bits/point +EBPFP 3.8952 equivalent bits/point +MSE 455.744194 +---------------------- -------------------------------------------------------- +Time: 0.342s Load: 0.003s, Pack+Encode: 0.166s, Decode+Unpack: 0.173s +---------------------- -------------------------------------------------------- +💾 Converting with 455.7442 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-467.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-467.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-470.zst (76/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-470.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 57, 128) +Output shape: (1, 57, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) -> torch.Size([1, 1, 57, 512]) + layer.4.output: torch.Size([1, 57, 3584]) -> torch.Size([1, 1, 57, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,636B, BPFP=0.9967 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,428B, BPFP=2.8586 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,124B, BPFP=1.6787 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,184B, BPFP=2.7917 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,996B, BPFP=1.9178 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,900B, BPFP=2.7138 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,500B, BPFP=1.7818 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,956B, BPFP=2.7292 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,820B, BPFP=2.4178 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,956B, BPFP=2.7292 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,752B, BPFP=1.0476 +⌛️ [2/4] FRONTEND: Frontend time: 0.156s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.173s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 57, 128]) + layer.0.v_cache: torch.Size([1, 4, 57, 128]) + layer.1.k_cache: torch.Size([1, 4, 57, 128]) + layer.1.v_cache: torch.Size([1, 4, 57, 128]) + layer.2.k_cache: torch.Size([1, 4, 57, 128]) + layer.2.v_cache: torch.Size([1, 4, 57, 128]) + layer.3.k_cache: torch.Size([1, 4, 57, 128]) + layer.3.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.k_cache: torch.Size([1, 4, 57, 128]) + layer.4.v_cache: torch.Size([1, 4, 57, 128]) + layer.4.output: torch.Size([1, 57, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10653571 31.92369963 + layer.0.v_cache 0.00001396 0.00560218 + layer.1.k_cache 0.01513677 4.10972836 + layer.1.v_cache 0.00000574 0.00227247 + layer.2.k_cache 0.00241650 0.53391878 + layer.2.v_cache 0.00001997 0.00572598 + layer.3.k_cache 0.04250543 2.26358086 + layer.3.v_cache 0.00001889 0.00665739 + layer.4.k_cache 0.00060099 0.12193073 + layer.4.v_cache 0.00005317 0.01372013 + layer.4.output 0.25650722 922.75916353 + ------------------------------------------------------------------------------------- + TOTAL 0.11546222 382.25299890 + (elements=496,128) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 496128 +Total Bytes 109252 +BPFP 1.7617 bits/point +EBPFP 3.5233 equivalent bits/point +MSE 382.252999 +---------------------- -------------------------------------------------------- +Time: 0.332s Load: 0.003s, Pack+Encode: 0.156s, Decode+Unpack: 0.173s +---------------------- -------------------------------------------------------- +💾 Converting with 382.2530 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-470.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-470.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-471.zst (77/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-471.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 48, 128) +Output shape: (1, 48, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) -> torch.Size([1, 1, 48, 512]) + layer.4.output: torch.Size([1, 48, 3584]) -> torch.Size([1, 1, 48, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,336B, BPFP=1.0859 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 9,640B, BPFP=3.1380 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,476B, BPFP=1.7826 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,180B, BPFP=2.9883 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,204B, BPFP=2.0195 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 8,992B, BPFP=2.9271 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 5,724B, BPFP=1.8633 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,152B, BPFP=2.9792 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,132B, BPFP=2.6471 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,192B, BPFP=2.9922 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,036B, BPFP=1.1642 +⌛️ [2/4] FRONTEND: Frontend time: 0.150s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.output: torch.Size([1, 48, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.172s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 48, 128]) + layer.0.v_cache: torch.Size([1, 4, 48, 128]) + layer.1.k_cache: torch.Size([1, 4, 48, 128]) + layer.1.v_cache: torch.Size([1, 4, 48, 128]) + layer.2.k_cache: torch.Size([1, 4, 48, 128]) + layer.2.v_cache: torch.Size([1, 4, 48, 128]) + layer.3.k_cache: torch.Size([1, 4, 48, 128]) + layer.3.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.k_cache: torch.Size([1, 4, 48, 128]) + layer.4.v_cache: torch.Size([1, 4, 48, 128]) + layer.4.output: torch.Size([1, 48, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10088900 37.67790476 + layer.0.v_cache 0.00001321 0.00543534 + layer.1.k_cache 0.01601078 3.93232918 + layer.1.v_cache 0.00000529 0.00211627 + layer.2.k_cache 0.00240732 0.50533561 + layer.2.v_cache 0.00001858 0.00636680 + layer.3.k_cache 0.04236705 2.78812408 + layer.3.v_cache 0.00001968 0.00662511 + layer.4.k_cache 0.00061965 0.12538057 + layer.4.v_cache 0.00005235 0.01443956 + layer.4.output 0.30498679 1054.97656250 + ------------------------------------------------------------------------------------- + TOTAL 0.13513591 437.05294087 + (elements=417,792) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 417792 +Total Bytes 100064 +BPFP 1.9161 bits/point +EBPFP 3.8321 equivalent bits/point +MSE 437.052941 +---------------------- -------------------------------------------------------- +Time: 0.324s Load: 0.003s, Pack+Encode: 0.150s, Decode+Unpack: 0.172s +---------------------- -------------------------------------------------------- +💾 Converting with 437.0529 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-471.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-471.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-496.zst (78/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-496.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 53, 128) +Output shape: (1, 53, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.0.v_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.1.k_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.1.v_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.2.k_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.2.v_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.3.k_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.3.v_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.4.k_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.4.v_cache: torch.Size([1, 4, 53, 128]) -> torch.Size([1, 1, 53, 512]) + layer.4.output: torch.Size([1, 53, 3584]) -> torch.Size([1, 1, 53, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,544B, BPFP=1.0448 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,328B, BPFP=3.0448 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,920B, BPFP=1.7453 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,108B, BPFP=2.9800 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,864B, BPFP=2.0236 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,840B, BPFP=2.9009 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,332B, BPFP=1.8667 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,896B, BPFP=2.9175 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,688B, BPFP=2.5613 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,696B, BPFP=2.8585 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 25,804B, BPFP=1.0868 +⌛️ [2/4] FRONTEND: Frontend time: 0.150s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 53, 128]) + layer.0.v_cache: torch.Size([1, 4, 53, 128]) + layer.1.k_cache: torch.Size([1, 4, 53, 128]) + layer.1.v_cache: torch.Size([1, 4, 53, 128]) + layer.2.k_cache: torch.Size([1, 4, 53, 128]) + layer.2.v_cache: torch.Size([1, 4, 53, 128]) + layer.3.k_cache: torch.Size([1, 4, 53, 128]) + layer.3.v_cache: torch.Size([1, 4, 53, 128]) + layer.4.k_cache: torch.Size([1, 4, 53, 128]) + layer.4.v_cache: torch.Size([1, 4, 53, 128]) + layer.4.output: torch.Size([1, 53, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.169s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 53, 128]) + layer.0.v_cache: torch.Size([1, 4, 53, 128]) + layer.1.k_cache: torch.Size([1, 4, 53, 128]) + layer.1.v_cache: torch.Size([1, 4, 53, 128]) + layer.2.k_cache: torch.Size([1, 4, 53, 128]) + layer.2.v_cache: torch.Size([1, 4, 53, 128]) + layer.3.k_cache: torch.Size([1, 4, 53, 128]) + layer.3.v_cache: torch.Size([1, 4, 53, 128]) + layer.4.k_cache: torch.Size([1, 4, 53, 128]) + layer.4.v_cache: torch.Size([1, 4, 53, 128]) + layer.4.output: torch.Size([1, 53, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10150213 32.94598504 + layer.0.v_cache 0.00001370 0.00613494 + layer.1.k_cache 0.01753690 3.80243611 + layer.1.v_cache 0.00000536 0.00258491 + layer.2.k_cache 0.00234856 0.51897700 + layer.2.v_cache 0.00001946 0.00713576 + layer.3.k_cache 0.03772523 2.49598089 + layer.3.v_cache 0.00001792 0.00739658 + layer.4.k_cache 0.00061365 0.13676455 + layer.4.v_cache 0.00006478 0.01531211 + layer.4.output 0.25806696 1009.49494609 + ------------------------------------------------------------------------------------- + TOTAL 0.11566567 418.02372533 + (elements=461,312) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 461312 +Total Bytes 107020 +BPFP 1.8559 bits/point +EBPFP 3.7118 equivalent bits/point +MSE 418.023725 +---------------------- -------------------------------------------------------- +Time: 0.322s Load: 0.003s, Pack+Encode: 0.150s, Decode+Unpack: 0.169s +---------------------- -------------------------------------------------------- +💾 Converting with 418.0237 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-496.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-496.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-498.zst (79/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-498.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 60, 128) +Output shape: (1, 60, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.output: torch.Size([1, 60, 3584]) -> torch.Size([1, 1, 60, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,948B, BPFP=1.0281 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,568B, BPFP=2.7521 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,436B, BPFP=1.6760 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,296B, BPFP=2.6812 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 7,280B, BPFP=1.8958 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 10,188B, BPFP=2.6531 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,832B, BPFP=1.7792 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 10,332B, BPFP=2.6906 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 9,052B, BPFP=2.3573 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 10,016B, BPFP=2.6083 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,364B, BPFP=1.0924 +⌛️ [2/4] FRONTEND: Frontend time: 0.162s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.output: torch.Size([1, 60, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.176s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.output: torch.Size([1, 60, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11850098 28.78559774 + layer.0.v_cache 0.00001385 0.00550965 + layer.1.k_cache 0.01468890 3.33212916 + layer.1.v_cache 0.00000523 0.00210399 + layer.2.k_cache 0.01477869 0.47933941 + layer.2.v_cache 0.00001915 0.00594608 + layer.3.k_cache 0.14661819 2.44952977 + layer.3.v_cache 0.00001925 0.00686226 + layer.4.k_cache 0.00060183 0.12412486 + layer.4.v_cache 0.00005752 0.01428895 + layer.4.output 0.22966646 876.34888393 + ------------------------------------------------------------------------------------- + TOTAL 0.11193934 362.92044820 + (elements=522,240) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 522240 +Total Bytes 114312 +BPFP 1.7511 bits/point +EBPFP 3.5022 equivalent bits/point +MSE 362.920448 +---------------------- -------------------------------------------------------- +Time: 0.341s Load: 0.003s, Pack+Encode: 0.162s, Decode+Unpack: 0.176s +---------------------- -------------------------------------------------------- +💾 Converting with 362.9204 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-498.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-498.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-504.zst (80/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-504.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 92, 128) +Output shape: (1, 92, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) -> torch.Size([1, 1, 92, 512]) + layer.4.output: torch.Size([1, 92, 3584]) -> torch.Size([1, 1, 92, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,548B, BPFP=0.9423 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,960B, BPFP=2.8804 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,784B, BPFP=1.6617 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,360B, BPFP=2.7785 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,836B, BPFP=1.8404 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,992B, BPFP=2.7160 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,180B, BPFP=1.7289 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,456B, BPFP=2.7948 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,380B, BPFP=2.4423 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,000B, BPFP=2.7174 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 40,328B, BPFP=0.9785 +⌛️ [2/4] FRONTEND: Frontend time: 0.268s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.326s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 92, 128]) + layer.0.v_cache: torch.Size([1, 4, 92, 128]) + layer.1.k_cache: torch.Size([1, 4, 92, 128]) + layer.1.v_cache: torch.Size([1, 4, 92, 128]) + layer.2.k_cache: torch.Size([1, 4, 92, 128]) + layer.2.v_cache: torch.Size([1, 4, 92, 128]) + layer.3.k_cache: torch.Size([1, 4, 92, 128]) + layer.3.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.k_cache: torch.Size([1, 4, 92, 128]) + layer.4.v_cache: torch.Size([1, 4, 92, 128]) + layer.4.output: torch.Size([1, 92, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10836371 29.68795644 + layer.0.v_cache 0.00001445 0.00510317 + layer.1.k_cache 0.06846340 3.40157484 + layer.1.v_cache 0.00000523 0.00183540 + layer.2.k_cache 0.00506382 0.47452823 + layer.2.v_cache 0.00001740 0.00483280 + layer.3.k_cache 0.03973729 2.60693708 + layer.3.v_cache 0.00001777 0.00573396 + layer.4.k_cache 0.00066797 0.11544603 + layer.4.v_cache 0.00005099 0.01193469 + layer.4.output 0.14786712 563.22287461 + ------------------------------------------------------------------------------------- + TOTAL 0.07396893 234.05152970 + (elements=800,768) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 800768 +Total Bytes 172824 +BPFP 1.7266 bits/point +EBPFP 3.4532 equivalent bits/point +MSE 234.051530 +---------------------- -------------------------------------------------------- +Time: 0.599s Load: 0.005s, Pack+Encode: 0.268s, Decode+Unpack: 0.326s +---------------------- -------------------------------------------------------- +💾 Converting with 234.0515 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-504.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-504.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-518.zst (81/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-518.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 66, 128) +Output shape: (1, 66, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.output: torch.Size([1, 66, 3584]) -> torch.Size([1, 1, 66, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,288B, BPFP=1.0152 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,236B, BPFP=3.3703 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 7,544B, BPFP=1.7860 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 12,936B, BPFP=3.0625 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,716B, BPFP=2.0634 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,376B, BPFP=2.9299 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 7,860B, BPFP=1.8608 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 12,928B, BPFP=3.0606 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,244B, BPFP=2.6619 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,020B, BPFP=3.0824 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 29,560B, BPFP=0.9997 +⌛️ [2/4] FRONTEND: Frontend time: 0.288s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.260s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08728886 33.92397239 + layer.0.v_cache 0.00001734 0.00530500 + layer.1.k_cache 0.01499850 3.59679274 + layer.1.v_cache 0.00000528 0.00193803 + layer.2.k_cache 0.00240521 0.48541445 + layer.2.v_cache 0.00001739 0.00530763 + layer.3.k_cache 0.03126823 2.28666502 + layer.3.v_cache 0.00001843 0.00626698 + layer.4.k_cache 0.00061395 0.12237803 + layer.4.v_cache 0.00004830 0.01160600 + layer.4.output 0.21953388 815.02746212 + ------------------------------------------------------------------------------------- + TOTAL 0.09843639 337.97869889 + (elements=574,464) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 574464 +Total Bytes 134708 +BPFP 1.8759 bits/point +EBPFP 3.7519 equivalent bits/point +MSE 337.978699 +---------------------- -------------------------------------------------------- +Time: 0.552s Load: 0.004s, Pack+Encode: 0.288s, Decode+Unpack: 0.260s +---------------------- -------------------------------------------------------- +💾 Converting with 337.9787 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-518.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-518.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-522.zst (82/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-522.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 66, 128) +Output shape: (1, 66, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) -> torch.Size([1, 1, 66, 512]) + layer.4.output: torch.Size([1, 66, 3584]) -> torch.Size([1, 1, 66, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,308B, BPFP=1.0199 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,336B, BPFP=3.3939 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 7,648B, BPFP=1.8106 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,084B, BPFP=3.0975 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 8,964B, BPFP=2.1222 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 12,844B, BPFP=3.0407 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,272B, BPFP=1.9583 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,256B, BPFP=3.1383 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,636B, BPFP=2.7547 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,436B, BPFP=3.1809 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,336B, BPFP=1.0260 +⌛️ [2/4] FRONTEND: Frontend time: 0.249s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.311s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 66, 128]) + layer.0.v_cache: torch.Size([1, 4, 66, 128]) + layer.1.k_cache: torch.Size([1, 4, 66, 128]) + layer.1.v_cache: torch.Size([1, 4, 66, 128]) + layer.2.k_cache: torch.Size([1, 4, 66, 128]) + layer.2.v_cache: torch.Size([1, 4, 66, 128]) + layer.3.k_cache: torch.Size([1, 4, 66, 128]) + layer.3.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.k_cache: torch.Size([1, 4, 66, 128]) + layer.4.v_cache: torch.Size([1, 4, 66, 128]) + layer.4.output: torch.Size([1, 66, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08223097 32.42665794 + layer.0.v_cache 0.00001442 0.00548488 + layer.1.k_cache 0.01480860 3.86250814 + layer.1.v_cache 0.00000502 0.00187894 + layer.2.k_cache 0.00576455 0.46963761 + layer.2.v_cache 0.00002304 0.00519455 + layer.3.k_cache 0.03215047 2.10663119 + layer.3.v_cache 0.00001933 0.00688280 + layer.4.k_cache 0.00061510 0.11743679 + layer.4.v_cache 0.00004916 0.01229996 + layer.4.output 0.22077460 810.82102273 + ------------------------------------------------------------------------------------- + TOTAL 0.09888840 336.16245717 + (elements=574,464) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 574464 +Total Bytes 138120 +BPFP 1.9235 bits/point +EBPFP 3.8469 equivalent bits/point +MSE 336.162457 +---------------------- -------------------------------------------------------- +Time: 0.564s Load: 0.003s, Pack+Encode: 0.249s, Decode+Unpack: 0.311s +---------------------- -------------------------------------------------------- +💾 Converting with 336.1625 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-522.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-522.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-523.zst (83/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-523.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 73, 128) +Output shape: (1, 73, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.output: torch.Size([1, 73, 3584]) -> torch.Size([1, 1, 73, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,840B, BPFP=1.0360 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,000B, BPFP=3.2106 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,224B, BPFP=1.7603 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,952B, BPFP=2.9863 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,396B, BPFP=2.0111 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,380B, BPFP=2.8639 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,744B, BPFP=1.8716 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,896B, BPFP=2.9743 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,244B, BPFP=2.6207 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,640B, BPFP=2.9195 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,228B, BPFP=0.9243 +⌛️ [2/4] FRONTEND: Frontend time: 0.275s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.output: torch.Size([1, 73, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.331s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.output: torch.Size([1, 73, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10114760 31.23528802 + layer.0.v_cache 0.00001420 0.00533300 + layer.1.k_cache 0.03683303 3.82064109 + layer.1.v_cache 0.00000544 0.00197738 + layer.2.k_cache 0.00248627 0.48454839 + layer.2.v_cache 0.00001713 0.00522181 + layer.3.k_cache 0.02909129 2.44785675 + layer.3.v_cache 0.00001766 0.00657601 + layer.4.k_cache 0.00072613 0.11999998 + layer.4.v_cache 0.00005154 0.01282451 + layer.4.output 0.19731793 736.22321429 + ------------------------------------------------------------------------------------- + TOTAL 0.09127152 305.39428041 + (elements=635,392) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 635392 +Total Bytes 143544 +BPFP 1.8073 bits/point +EBPFP 3.6146 equivalent bits/point +MSE 305.394280 +---------------------- -------------------------------------------------------- +Time: 0.610s Load: 0.004s, Pack+Encode: 0.275s, Decode+Unpack: 0.331s +---------------------- -------------------------------------------------------- +💾 Converting with 305.3943 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-523.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-523.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-531.zst (84/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-531.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 76, 128) +Output shape: (1, 76, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) -> torch.Size([1, 1, 76, 512]) + layer.4.output: torch.Size([1, 76, 3584]) -> torch.Size([1, 1, 76, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,004B, BPFP=1.0288 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,448B, BPFP=3.1760 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,400B, BPFP=1.7270 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,556B, BPFP=2.9926 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,688B, BPFP=1.9918 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,084B, BPFP=2.8956 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,016B, BPFP=1.8536 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,624B, BPFP=3.0066 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,656B, BPFP=2.6020 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,188B, BPFP=2.9169 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,092B, BPFP=0.9719 +⌛️ [2/4] FRONTEND: Frontend time: 0.262s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.output: torch.Size([1, 76, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.341s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 76, 128]) + layer.0.v_cache: torch.Size([1, 4, 76, 128]) + layer.1.k_cache: torch.Size([1, 4, 76, 128]) + layer.1.v_cache: torch.Size([1, 4, 76, 128]) + layer.2.k_cache: torch.Size([1, 4, 76, 128]) + layer.2.v_cache: torch.Size([1, 4, 76, 128]) + layer.3.k_cache: torch.Size([1, 4, 76, 128]) + layer.3.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.k_cache: torch.Size([1, 4, 76, 128]) + layer.4.v_cache: torch.Size([1, 4, 76, 128]) + layer.4.output: torch.Size([1, 76, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10230959 29.72106612 + layer.0.v_cache 0.00001409 0.00532505 + layer.1.k_cache 0.07973289 3.81154713 + layer.1.v_cache 0.00000506 0.00178405 + layer.2.k_cache 0.00553133 0.46045213 + layer.2.v_cache 0.00001928 0.00500714 + layer.3.k_cache 0.03946657 2.22874531 + layer.3.v_cache 0.00001980 0.00611967 + layer.4.k_cache 0.00066525 0.11894579 + layer.4.v_cache 0.00005119 0.01223754 + layer.4.output 0.17985767 690.47174577 + ------------------------------------------------------------------------------------- + TOTAL 0.08745993 286.45137943 + (elements=661,504) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 661504 +Total Bytes 150756 +BPFP 1.8232 bits/point +EBPFP 3.6464 equivalent bits/point +MSE 286.451379 +---------------------- -------------------------------------------------------- +Time: 0.607s Load: 0.004s, Pack+Encode: 0.262s, Decode+Unpack: 0.341s +---------------------- -------------------------------------------------------- +💾 Converting with 286.4514 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-531.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-531.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-533.zst (85/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-533.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 77, 128) +Output shape: (1, 77, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) -> torch.Size([1, 1, 77, 512]) + layer.4.output: torch.Size([1, 77, 3584]) -> torch.Size([1, 1, 77, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,984B, BPFP=1.0114 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,996B, BPFP=3.0430 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,232B, BPFP=1.6705 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,056B, BPFP=2.8523 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,268B, BPFP=1.8807 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,720B, BPFP=2.7841 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,572B, BPFP=1.7394 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,184B, BPFP=2.8782 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,268B, BPFP=2.4894 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,888B, BPFP=2.8182 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,312B, BPFP=0.8787 +⌛️ [2/4] FRONTEND: Frontend time: 0.257s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.286s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 77, 128]) + layer.0.v_cache: torch.Size([1, 4, 77, 128]) + layer.1.k_cache: torch.Size([1, 4, 77, 128]) + layer.1.v_cache: torch.Size([1, 4, 77, 128]) + layer.2.k_cache: torch.Size([1, 4, 77, 128]) + layer.2.v_cache: torch.Size([1, 4, 77, 128]) + layer.3.k_cache: torch.Size([1, 4, 77, 128]) + layer.3.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.k_cache: torch.Size([1, 4, 77, 128]) + layer.4.v_cache: torch.Size([1, 4, 77, 128]) + layer.4.output: torch.Size([1, 77, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10633976 31.27620865 + layer.0.v_cache 0.00001461 0.00520501 + layer.1.k_cache 0.07748085 3.20710853 + layer.1.v_cache 0.00000515 0.00171715 + layer.2.k_cache 0.00505460 0.53868509 + layer.2.v_cache 0.00001651 0.00491550 + layer.3.k_cache 0.02793909 2.52609312 + layer.3.v_cache 0.00001809 0.00567124 + layer.4.k_cache 0.00070819 0.12035400 + layer.4.v_cache 0.00004716 0.01154297 + layer.4.output 0.18765952 698.92584647 + ------------------------------------------------------------------------------------- + TOTAL 0.09007298 290.01049568 + (elements=670,208) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 670208 +Total Bytes 144480 +BPFP 1.7246 bits/point +EBPFP 3.4492 equivalent bits/point +MSE 290.010496 +---------------------- -------------------------------------------------------- +Time: 0.547s Load: 0.004s, Pack+Encode: 0.257s, Decode+Unpack: 0.286s +---------------------- -------------------------------------------------------- +💾 Converting with 290.0105 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-533.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-533.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-537.zst (86/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-537.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 70, 128) +Output shape: (1, 70, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) -> torch.Size([1, 1, 70, 512]) + layer.4.output: torch.Size([1, 70, 3584]) -> torch.Size([1, 1, 70, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,620B, BPFP=1.0312 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,696B, BPFP=3.2804 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,100B, BPFP=1.8080 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,704B, BPFP=3.0589 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,276B, BPFP=2.0705 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,216B, BPFP=2.9500 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,600B, BPFP=1.9196 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,812B, BPFP=3.0830 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,848B, BPFP=2.6446 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,392B, BPFP=2.9893 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,612B, BPFP=0.9124 +⌛️ [2/4] FRONTEND: Frontend time: 0.278s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.301s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 70, 128]) + layer.0.v_cache: torch.Size([1, 4, 70, 128]) + layer.1.k_cache: torch.Size([1, 4, 70, 128]) + layer.1.v_cache: torch.Size([1, 4, 70, 128]) + layer.2.k_cache: torch.Size([1, 4, 70, 128]) + layer.2.v_cache: torch.Size([1, 4, 70, 128]) + layer.3.k_cache: torch.Size([1, 4, 70, 128]) + layer.3.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.k_cache: torch.Size([1, 4, 70, 128]) + layer.4.v_cache: torch.Size([1, 4, 70, 128]) + layer.4.output: torch.Size([1, 70, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08399124 31.18534459 + layer.0.v_cache 0.00001671 0.00517412 + layer.1.k_cache 0.03849935 3.77927595 + layer.1.v_cache 0.00000503 0.00178516 + layer.2.k_cache 0.00738276 0.47550681 + layer.2.v_cache 0.00001692 0.00474117 + layer.3.k_cache 0.04914829 2.26442610 + layer.3.v_cache 0.00001713 0.00580190 + layer.4.k_cache 0.00063731 0.11362971 + layer.4.v_cache 0.00004711 0.01151480 + layer.4.output 0.21328995 765.15542092 + ------------------------------------------------------------------------------------- + TOTAL 0.09839950 317.29030275 + (elements=609,280) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 609280 +Total Bytes 139876 +BPFP 1.8366 bits/point +EBPFP 3.6732 equivalent bits/point +MSE 317.290303 +---------------------- -------------------------------------------------------- +Time: 0.583s Load: 0.004s, Pack+Encode: 0.278s, Decode+Unpack: 0.301s +---------------------- -------------------------------------------------------- +💾 Converting with 317.2903 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-537.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-537.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-54.zst (87/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-54.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 56, 128) +Output shape: (1, 56, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.0.v_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.1.k_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.1.v_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.2.k_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.2.v_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.3.k_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.3.v_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.4.k_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.4.v_cache: torch.Size([1, 4, 56, 128]) -> torch.Size([1, 1, 56, 512]) + layer.4.output: torch.Size([1, 56, 3584]) -> torch.Size([1, 1, 56, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,600B, BPFP=1.0045 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,200B, BPFP=2.8460 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 5,904B, BPFP=1.6473 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 9,848B, BPFP=2.7478 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 6,648B, BPFP=1.8549 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,700B, BPFP=2.7065 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,252B, BPFP=1.7444 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 9,860B, BPFP=2.7511 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,640B, BPFP=2.4107 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,660B, BPFP=2.6953 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 26,388B, BPFP=1.0518 +⌛️ [2/4] FRONTEND: Frontend time: 0.152s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 56, 128]) + layer.0.v_cache: torch.Size([1, 4, 56, 128]) + layer.1.k_cache: torch.Size([1, 4, 56, 128]) + layer.1.v_cache: torch.Size([1, 4, 56, 128]) + layer.2.k_cache: torch.Size([1, 4, 56, 128]) + layer.2.v_cache: torch.Size([1, 4, 56, 128]) + layer.3.k_cache: torch.Size([1, 4, 56, 128]) + layer.3.v_cache: torch.Size([1, 4, 56, 128]) + layer.4.k_cache: torch.Size([1, 4, 56, 128]) + layer.4.v_cache: torch.Size([1, 4, 56, 128]) + layer.4.output: torch.Size([1, 56, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.177s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 56, 128]) + layer.0.v_cache: torch.Size([1, 4, 56, 128]) + layer.1.k_cache: torch.Size([1, 4, 56, 128]) + layer.1.v_cache: torch.Size([1, 4, 56, 128]) + layer.2.k_cache: torch.Size([1, 4, 56, 128]) + layer.2.v_cache: torch.Size([1, 4, 56, 128]) + layer.3.k_cache: torch.Size([1, 4, 56, 128]) + layer.3.v_cache: torch.Size([1, 4, 56, 128]) + layer.4.k_cache: torch.Size([1, 4, 56, 128]) + layer.4.v_cache: torch.Size([1, 4, 56, 128]) + layer.4.output: torch.Size([1, 56, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13768739 32.30942208 + layer.0.v_cache 0.00001432 0.00566723 + layer.1.k_cache 0.01638686 3.51667023 + layer.1.v_cache 0.00000533 0.00208194 + layer.2.k_cache 0.00251956 0.46410952 + layer.2.v_cache 0.00001745 0.00571134 + layer.3.k_cache 0.08599430 2.48888070 + layer.3.v_cache 0.00002012 0.00749257 + layer.4.k_cache 0.00061995 0.12364937 + layer.4.v_cache 0.00004798 0.01263006 + layer.4.output 0.24256272 944.71197385 + ------------------------------------------------------------------------------------- + TOTAL 0.11419131 391.28941953 + (elements=487,424) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 487424 +Total Bytes 106700 +BPFP 1.7512 bits/point +EBPFP 3.5025 equivalent bits/point +MSE 391.289420 +---------------------- -------------------------------------------------------- +Time: 0.332s Load: 0.003s, Pack+Encode: 0.152s, Decode+Unpack: 0.177s +---------------------- -------------------------------------------------------- +💾 Converting with 391.2894 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-54.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-54.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-540.zst (88/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-540.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 80, 128) +Output shape: (1, 80, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) -> torch.Size([1, 1, 80, 512]) + layer.4.output: torch.Size([1, 80, 3584]) -> torch.Size([1, 1, 80, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,940B, BPFP=0.9648 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,584B, BPFP=3.0438 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,528B, BPFP=1.6656 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,892B, BPFP=2.9086 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,564B, BPFP=1.8680 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,448B, BPFP=2.8219 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,028B, BPFP=1.7633 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,808B, BPFP=2.8922 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,888B, BPFP=2.5172 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 14,480B, BPFP=2.8281 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,720B, BPFP=0.9408 +⌛️ [2/4] FRONTEND: Frontend time: 0.275s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.245s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 80, 128]) + layer.0.v_cache: torch.Size([1, 4, 80, 128]) + layer.1.k_cache: torch.Size([1, 4, 80, 128]) + layer.1.v_cache: torch.Size([1, 4, 80, 128]) + layer.2.k_cache: torch.Size([1, 4, 80, 128]) + layer.2.v_cache: torch.Size([1, 4, 80, 128]) + layer.3.k_cache: torch.Size([1, 4, 80, 128]) + layer.3.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.k_cache: torch.Size([1, 4, 80, 128]) + layer.4.v_cache: torch.Size([1, 4, 80, 128]) + layer.4.output: torch.Size([1, 80, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09278157 30.78129272 + layer.0.v_cache 0.00001504 0.00505186 + layer.1.k_cache 0.07523044 3.83266602 + layer.1.v_cache 0.00000517 0.00187673 + layer.2.k_cache 0.00381728 0.45607872 + layer.2.v_cache 0.00001789 0.00514870 + layer.3.k_cache 0.03327554 2.25985107 + layer.3.v_cache 0.00001939 0.00587529 + layer.4.k_cache 0.00067710 0.11898481 + layer.4.v_cache 0.00005192 0.01216211 + layer.4.output 0.16994183 664.91428571 + ------------------------------------------------------------------------------------- + TOTAL 0.08208730 275.99288165 + (elements=696,320) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 696320 +Total Bytes 152880 +BPFP 1.7564 bits/point +EBPFP 3.5129 equivalent bits/point +MSE 275.992882 +---------------------- -------------------------------------------------------- +Time: 0.524s Load: 0.004s, Pack+Encode: 0.275s, Decode+Unpack: 0.245s +---------------------- -------------------------------------------------------- +💾 Converting with 275.9929 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-540.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-540.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-57.zst (89/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-57.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 59, 128) +Output shape: (1, 59, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) -> torch.Size([1, 1, 59, 512]) + layer.4.output: torch.Size([1, 59, 3584]) -> torch.Size([1, 1, 59, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,848B, BPFP=1.0191 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,524B, BPFP=2.7871 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,264B, BPFP=1.6589 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,172B, BPFP=2.6939 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 7,056B, BPFP=1.8686 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 9,904B, BPFP=2.6229 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,660B, BPFP=1.7638 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 10,208B, BPFP=2.7034 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 8,884B, BPFP=2.3528 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 9,860B, BPFP=2.6112 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,088B, BPFP=1.0627 +⌛️ [2/4] FRONTEND: Frontend time: 0.152s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.output: torch.Size([1, 59, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.174s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 59, 128]) + layer.0.v_cache: torch.Size([1, 4, 59, 128]) + layer.1.k_cache: torch.Size([1, 4, 59, 128]) + layer.1.v_cache: torch.Size([1, 4, 59, 128]) + layer.2.k_cache: torch.Size([1, 4, 59, 128]) + layer.2.v_cache: torch.Size([1, 4, 59, 128]) + layer.3.k_cache: torch.Size([1, 4, 59, 128]) + layer.3.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.k_cache: torch.Size([1, 4, 59, 128]) + layer.4.v_cache: torch.Size([1, 4, 59, 128]) + layer.4.output: torch.Size([1, 59, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12468526 28.63478425 + layer.0.v_cache 0.00001451 0.00585177 + layer.1.k_cache 0.01313243 3.65344730 + layer.1.v_cache 0.00000580 0.00217278 + layer.2.k_cache 0.00906488 0.46179868 + layer.2.v_cache 0.00002099 0.00595002 + layer.3.k_cache 0.01569995 2.44398214 + layer.3.v_cache 0.00002002 0.00695885 + layer.4.k_cache 0.00064964 0.11834864 + layer.4.v_cache 0.00005478 0.01277868 + layer.4.output 0.23031524 888.05894370 + ------------------------------------------------------------------------------------- + TOTAL 0.10444441 367.75051053 + (elements=513,536) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 513536 +Total Bytes 111468 +BPFP 1.7365 bits/point +EBPFP 3.4730 equivalent bits/point +MSE 367.750511 +---------------------- -------------------------------------------------------- +Time: 0.329s Load: 0.004s, Pack+Encode: 0.152s, Decode+Unpack: 0.174s +---------------------- -------------------------------------------------------- +💾 Converting with 367.7505 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-57.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-57.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-59.zst (90/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-59.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.003s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 60, 128) +Output shape: (1, 60, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) -> torch.Size([1, 1, 60, 512]) + layer.4.output: torch.Size([1, 60, 3584]) -> torch.Size([1, 1, 60, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 3,968B, BPFP=1.0333 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 10,612B, BPFP=2.7635 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 6,408B, BPFP=1.6687 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 10,296B, BPFP=2.6812 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 7,208B, BPFP=1.8771 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 10,152B, BPFP=2.6437 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 6,784B, BPFP=1.7667 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 10,268B, BPFP=2.6740 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 9,016B, BPFP=2.3479 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 10,076B, BPFP=2.6240 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 30,508B, BPFP=1.1350 +⌛️ [2/4] FRONTEND: Frontend time: 0.157s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.output: torch.Size([1, 60, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.178s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 60, 128]) + layer.0.v_cache: torch.Size([1, 4, 60, 128]) + layer.1.k_cache: torch.Size([1, 4, 60, 128]) + layer.1.v_cache: torch.Size([1, 4, 60, 128]) + layer.2.k_cache: torch.Size([1, 4, 60, 128]) + layer.2.v_cache: torch.Size([1, 4, 60, 128]) + layer.3.k_cache: torch.Size([1, 4, 60, 128]) + layer.3.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.k_cache: torch.Size([1, 4, 60, 128]) + layer.4.v_cache: torch.Size([1, 4, 60, 128]) + layer.4.output: torch.Size([1, 60, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.15164433 30.43490397 + layer.0.v_cache 0.00001435 0.00597720 + layer.1.k_cache 0.01704682 3.90348104 + layer.1.v_cache 0.00000573 0.00236019 + layer.2.k_cache 0.00240009 0.50211693 + layer.2.v_cache 0.00001845 0.00563602 + layer.3.k_cache 0.02161660 2.44555537 + layer.3.v_cache 0.00001985 0.00666304 + layer.4.k_cache 0.00061459 0.12494114 + layer.4.v_cache 0.00005275 0.01386226 + layer.4.output 0.22653077 871.22752976 + ------------------------------------------------------------------------------------- + TOTAL 0.10465582 360.94342385 + (elements=522,240) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 522240 +Total Bytes 115296 +BPFP 1.7662 bits/point +EBPFP 3.5324 equivalent bits/point +MSE 360.943424 +---------------------- -------------------------------------------------------- +Time: 0.338s Load: 0.003s, Pack+Encode: 0.157s, Decode+Unpack: 0.178s +---------------------- -------------------------------------------------------- +💾 Converting with 360.9434 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-59.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-59.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-63.zst (91/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-63.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 73, 128) +Output shape: (1, 73, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) -> torch.Size([1, 1, 73, 512]) + layer.4.output: torch.Size([1, 73, 3584]) -> torch.Size([1, 1, 73, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,848B, BPFP=1.0377 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,512B, BPFP=3.1062 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,024B, BPFP=1.7175 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,676B, BPFP=2.9272 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,052B, BPFP=1.9375 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,224B, BPFP=2.8305 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,548B, BPFP=1.8296 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,580B, BPFP=2.9067 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,128B, BPFP=2.5959 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,380B, BPFP=2.8639 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 27,932B, BPFP=0.8541 +⌛️ [2/4] FRONTEND: Frontend time: 0.261s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.output: torch.Size([1, 73, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.325s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 73, 128]) + layer.0.v_cache: torch.Size([1, 4, 73, 128]) + layer.1.k_cache: torch.Size([1, 4, 73, 128]) + layer.1.v_cache: torch.Size([1, 4, 73, 128]) + layer.2.k_cache: torch.Size([1, 4, 73, 128]) + layer.2.v_cache: torch.Size([1, 4, 73, 128]) + layer.3.k_cache: torch.Size([1, 4, 73, 128]) + layer.3.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.k_cache: torch.Size([1, 4, 73, 128]) + layer.4.v_cache: torch.Size([1, 4, 73, 128]) + layer.4.output: torch.Size([1, 73, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.08869729 31.58395762 + layer.0.v_cache 0.00001594 0.00495882 + layer.1.k_cache 0.08105310 3.68917303 + layer.1.v_cache 0.00000494 0.00169014 + layer.2.k_cache 0.00256802 0.51804843 + layer.2.v_cache 0.00001661 0.00447366 + layer.3.k_cache 0.01835501 2.47350531 + layer.3.v_cache 0.00001861 0.00555920 + layer.4.k_cache 0.00078324 0.11957722 + layer.4.v_cache 0.00004769 0.01092144 + layer.4.output 0.18611241 680.51632828 + ------------------------------------------------------------------------------------- + TOTAL 0.08790278 282.47212722 + (elements=635,392) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 635392 +Total Bytes 138904 +BPFP 1.7489 bits/point +EBPFP 3.4978 equivalent bits/point +MSE 282.472127 +---------------------- -------------------------------------------------------- +Time: 0.591s Load: 0.004s, Pack+Encode: 0.261s, Decode+Unpack: 0.325s +---------------------- -------------------------------------------------------- +💾 Converting with 282.4721 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-63.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-63.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-65.zst (92/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-65.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 97, 128) +Output shape: (1, 97, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) -> torch.Size([1, 1, 97, 512]) + layer.4.output: torch.Size([1, 97, 3584]) -> torch.Size([1, 1, 97, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,212B, BPFP=0.8396 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,368B, BPFP=2.7977 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,536B, BPFP=1.5361 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,564B, BPFP=2.6682 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,656B, BPFP=1.7165 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,188B, BPFP=2.6076 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,180B, BPFP=1.6398 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,612B, BPFP=2.6759 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,344B, BPFP=2.3106 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,004B, BPFP=2.5780 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 38,500B, BPFP=0.8860 +⌛️ [2/4] FRONTEND: Frontend time: 0.243s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.output: torch.Size([1, 97, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.293s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 97, 128]) + layer.0.v_cache: torch.Size([1, 4, 97, 128]) + layer.1.k_cache: torch.Size([1, 4, 97, 128]) + layer.1.v_cache: torch.Size([1, 4, 97, 128]) + layer.2.k_cache: torch.Size([1, 4, 97, 128]) + layer.2.v_cache: torch.Size([1, 4, 97, 128]) + layer.3.k_cache: torch.Size([1, 4, 97, 128]) + layer.3.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.k_cache: torch.Size([1, 4, 97, 128]) + layer.4.v_cache: torch.Size([1, 4, 97, 128]) + layer.4.output: torch.Size([1, 97, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.17228427 30.06260319 + layer.0.v_cache 0.00001543 0.00491311 + layer.1.k_cache 0.04621154 3.42836234 + layer.1.v_cache 0.00000545 0.00170948 + layer.2.k_cache 0.00739691 0.42975357 + layer.2.v_cache 0.00001797 0.00461751 + layer.3.k_cache 0.03096496 2.21281559 + layer.3.v_cache 0.00001890 0.00591365 + layer.4.k_cache 0.00071091 0.11666084 + layer.4.v_cache 0.00004900 0.01041454 + layer.4.output 0.02450962 557.79906112 + ------------------------------------------------------------------------------------- + TOTAL 0.02524957 231.81595245 + (elements=844,288) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 844288 +Total Bytes 171164 +BPFP 1.6219 bits/point +EBPFP 3.2437 equivalent bits/point +MSE 231.815952 +---------------------- -------------------------------------------------------- +Time: 0.540s Load: 0.004s, Pack+Encode: 0.243s, Decode+Unpack: 0.293s +---------------------- -------------------------------------------------------- +💾 Converting with 231.8160 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-65.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-65.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-67.zst (93/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-67.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 93, 128) +Output shape: (1, 93, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.0.v_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.1.k_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.1.v_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.2.k_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.2.v_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.3.k_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.3.v_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.4.k_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.4.v_cache: torch.Size([1, 4, 93, 128]) -> torch.Size([1, 1, 93, 512]) + layer.4.output: torch.Size([1, 93, 3584]) -> torch.Size([1, 1, 93, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,644B, BPFP=0.9483 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 17,292B, BPFP=2.9052 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,972B, BPFP=1.6754 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 16,832B, BPFP=2.8280 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,876B, BPFP=1.8273 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 16,376B, BPFP=2.7513 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,340B, BPFP=1.7372 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 16,684B, BPFP=2.8031 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 14,580B, BPFP=2.4496 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 16,236B, BPFP=2.7278 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 38,636B, BPFP=0.9273 +⌛️ [2/4] FRONTEND: Frontend time: 0.246s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 93, 128]) + layer.0.v_cache: torch.Size([1, 4, 93, 128]) + layer.1.k_cache: torch.Size([1, 4, 93, 128]) + layer.1.v_cache: torch.Size([1, 4, 93, 128]) + layer.2.k_cache: torch.Size([1, 4, 93, 128]) + layer.2.v_cache: torch.Size([1, 4, 93, 128]) + layer.3.k_cache: torch.Size([1, 4, 93, 128]) + layer.3.v_cache: torch.Size([1, 4, 93, 128]) + layer.4.k_cache: torch.Size([1, 4, 93, 128]) + layer.4.v_cache: torch.Size([1, 4, 93, 128]) + layer.4.output: torch.Size([1, 93, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.286s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 93, 128]) + layer.0.v_cache: torch.Size([1, 4, 93, 128]) + layer.1.k_cache: torch.Size([1, 4, 93, 128]) + layer.1.v_cache: torch.Size([1, 4, 93, 128]) + layer.2.k_cache: torch.Size([1, 4, 93, 128]) + layer.2.v_cache: torch.Size([1, 4, 93, 128]) + layer.3.k_cache: torch.Size([1, 4, 93, 128]) + layer.3.v_cache: torch.Size([1, 4, 93, 128]) + layer.4.k_cache: torch.Size([1, 4, 93, 128]) + layer.4.v_cache: torch.Size([1, 4, 93, 128]) + layer.4.output: torch.Size([1, 93, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.16174046 29.04042759 + layer.0.v_cache 0.00001612 0.00541099 + layer.1.k_cache 0.03352992 3.67366536 + layer.1.v_cache 0.00000563 0.00201894 + layer.2.k_cache 0.00496730 0.48507727 + layer.2.v_cache 0.00001740 0.00498876 + layer.3.k_cache 0.01521110 2.38934064 + layer.3.v_cache 0.00001939 0.00619674 + layer.4.k_cache 0.00066431 0.12252608 + layer.4.v_cache 0.00005067 0.01212539 + layer.4.output 0.14624557 578.55712366 + ------------------------------------------------------------------------------------- + TOTAL 0.07293772 240.33186137 + (elements=809,472) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 809472 +Total Bytes 173468 +BPFP 1.7144 bits/point +EBPFP 3.4288 equivalent bits/point +MSE 240.331861 +---------------------- -------------------------------------------------------- +Time: 0.537s Load: 0.004s, Pack+Encode: 0.246s, Decode+Unpack: 0.286s +---------------------- -------------------------------------------------------- +💾 Converting with 240.3319 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-67.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-67.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-7.zst (94/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-7.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 86, 128) +Output shape: (1, 86, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.0.v_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.1.k_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.1.v_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.2.k_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.2.v_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.3.k_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.3.v_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.4.k_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.4.v_cache: torch.Size([1, 4, 86, 128]) -> torch.Size([1, 1, 86, 512]) + layer.4.output: torch.Size([1, 86, 3584]) -> torch.Size([1, 1, 86, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,176B, BPFP=0.9404 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,144B, BPFP=2.9331 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,768B, BPFP=1.5930 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,432B, BPFP=2.8038 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,016B, BPFP=1.8198 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 14,964B, BPFP=2.7188 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,472B, BPFP=1.7209 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,500B, BPFP=2.8161 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,288B, BPFP=2.4142 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,020B, BPFP=2.7289 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 33,828B, BPFP=0.8780 +⌛️ [2/4] FRONTEND: Frontend time: 0.252s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 86, 128]) + layer.0.v_cache: torch.Size([1, 4, 86, 128]) + layer.1.k_cache: torch.Size([1, 4, 86, 128]) + layer.1.v_cache: torch.Size([1, 4, 86, 128]) + layer.2.k_cache: torch.Size([1, 4, 86, 128]) + layer.2.v_cache: torch.Size([1, 4, 86, 128]) + layer.3.k_cache: torch.Size([1, 4, 86, 128]) + layer.3.v_cache: torch.Size([1, 4, 86, 128]) + layer.4.k_cache: torch.Size([1, 4, 86, 128]) + layer.4.v_cache: torch.Size([1, 4, 86, 128]) + layer.4.output: torch.Size([1, 86, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.292s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 86, 128]) + layer.0.v_cache: torch.Size([1, 4, 86, 128]) + layer.1.k_cache: torch.Size([1, 4, 86, 128]) + layer.1.v_cache: torch.Size([1, 4, 86, 128]) + layer.2.k_cache: torch.Size([1, 4, 86, 128]) + layer.2.v_cache: torch.Size([1, 4, 86, 128]) + layer.3.k_cache: torch.Size([1, 4, 86, 128]) + layer.3.v_cache: torch.Size([1, 4, 86, 128]) + layer.4.k_cache: torch.Size([1, 4, 86, 128]) + layer.4.v_cache: torch.Size([1, 4, 86, 128]) + layer.4.output: torch.Size([1, 86, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.07121617 28.34774142 + layer.0.v_cache 0.00001681 0.00483551 + layer.1.k_cache 0.06877367 3.58687042 + layer.1.v_cache 0.00000530 0.00167125 + layer.2.k_cache 0.00486504 0.44219305 + layer.2.v_cache 0.00001749 0.00421199 + layer.3.k_cache 0.08833314 1.99296712 + layer.3.v_cache 0.00001723 0.00543816 + layer.4.k_cache 0.00077278 0.10893376 + layer.4.v_cache 0.00004595 0.01075802 + layer.4.output 0.15812202 625.15765158 + ------------------------------------------------------------------------------------- + TOTAL 0.07887751 259.44759893 + (elements=748,544) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 748544 +Total Bytes 157608 +BPFP 1.6844 bits/point +EBPFP 3.3688 equivalent bits/point +MSE 259.447599 +---------------------- -------------------------------------------------------- +Time: 0.548s Load: 0.004s, Pack+Encode: 0.252s, Decode+Unpack: 0.292s +---------------------- -------------------------------------------------------- +💾 Converting with 259.4476 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-7.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-7.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-72.zst (95/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-72.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 101, 128) +Output shape: (1, 101, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.0.v_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.1.k_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.1.v_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.2.k_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.2.v_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.3.k_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.3.v_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.4.k_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.4.v_cache: torch.Size([1, 4, 101, 128]) -> torch.Size([1, 1, 101, 512]) + layer.4.output: torch.Size([1, 101, 3584]) -> torch.Size([1, 1, 101, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,816B, BPFP=0.8998 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 18,540B, BPFP=2.8682 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,368B, BPFP=1.6040 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 17,880B, BPFP=2.7661 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,312B, BPFP=1.7500 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 17,384B, BPFP=2.6894 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 10,960B, BPFP=1.6955 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 17,792B, BPFP=2.7525 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 15,448B, BPFP=2.3899 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 17,280B, BPFP=2.6733 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 36,980B, BPFP=0.8173 +⌛️ [2/4] FRONTEND: Frontend time: 0.253s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 101, 128]) + layer.0.v_cache: torch.Size([1, 4, 101, 128]) + layer.1.k_cache: torch.Size([1, 4, 101, 128]) + layer.1.v_cache: torch.Size([1, 4, 101, 128]) + layer.2.k_cache: torch.Size([1, 4, 101, 128]) + layer.2.v_cache: torch.Size([1, 4, 101, 128]) + layer.3.k_cache: torch.Size([1, 4, 101, 128]) + layer.3.v_cache: torch.Size([1, 4, 101, 128]) + layer.4.k_cache: torch.Size([1, 4, 101, 128]) + layer.4.v_cache: torch.Size([1, 4, 101, 128]) + layer.4.output: torch.Size([1, 101, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.295s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 101, 128]) + layer.0.v_cache: torch.Size([1, 4, 101, 128]) + layer.1.k_cache: torch.Size([1, 4, 101, 128]) + layer.1.v_cache: torch.Size([1, 4, 101, 128]) + layer.2.k_cache: torch.Size([1, 4, 101, 128]) + layer.2.v_cache: torch.Size([1, 4, 101, 128]) + layer.3.k_cache: torch.Size([1, 4, 101, 128]) + layer.3.v_cache: torch.Size([1, 4, 101, 128]) + layer.4.k_cache: torch.Size([1, 4, 101, 128]) + layer.4.v_cache: torch.Size([1, 4, 101, 128]) + layer.4.output: torch.Size([1, 101, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.13964315 28.34033445 + layer.0.v_cache 0.00001432 0.00533649 + layer.1.k_cache 0.04810369 3.39681795 + layer.1.v_cache 0.00000572 0.00207072 + layer.2.k_cache 0.00492308 0.47415308 + layer.2.v_cache 0.00001777 0.00515716 + layer.3.k_cache 0.04255116 2.31087683 + layer.3.v_cache 0.00001880 0.00612267 + layer.4.k_cache 0.00073971 0.12119390 + layer.4.v_cache 0.00004573 0.01214448 + layer.4.output 11.28739116 527.86558522 + ------------------------------------------------------------------------------------- + TOTAL 4.66163537 219.39607672 + (elements=879,104) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 879104 +Total Bytes 179760 +BPFP 1.6358 bits/point +EBPFP 3.2717 equivalent bits/point +MSE 219.396077 +---------------------- -------------------------------------------------------- +Time: 0.553s Load: 0.005s, Pack+Encode: 0.253s, Decode+Unpack: 0.295s +---------------------- -------------------------------------------------------- +💾 Converting with 219.3961 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-72.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-72.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-75.zst (96/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-75.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 88, 128) +Output shape: (1, 88, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) -> torch.Size([1, 1, 88, 512]) + layer.4.output: torch.Size([1, 88, 3584]) -> torch.Size([1, 1, 88, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,308B, BPFP=0.9425 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 16,324B, BPFP=2.8984 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 9,040B, BPFP=1.6051 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 15,608B, BPFP=2.7713 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 10,212B, BPFP=1.8132 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 15,260B, BPFP=2.7095 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 9,632B, BPFP=1.7102 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 15,592B, BPFP=2.7685 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 13,656B, BPFP=2.4247 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 15,152B, BPFP=2.6903 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 34,216B, BPFP=0.8679 +⌛️ [2/4] FRONTEND: Frontend time: 0.227s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.324s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 88, 128]) + layer.0.v_cache: torch.Size([1, 4, 88, 128]) + layer.1.k_cache: torch.Size([1, 4, 88, 128]) + layer.1.v_cache: torch.Size([1, 4, 88, 128]) + layer.2.k_cache: torch.Size([1, 4, 88, 128]) + layer.2.v_cache: torch.Size([1, 4, 88, 128]) + layer.3.k_cache: torch.Size([1, 4, 88, 128]) + layer.3.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.k_cache: torch.Size([1, 4, 88, 128]) + layer.4.v_cache: torch.Size([1, 4, 88, 128]) + layer.4.output: torch.Size([1, 88, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.14145413 28.05453214 + layer.0.v_cache 0.00001516 0.00492600 + layer.1.k_cache 0.10647225 3.45957531 + layer.1.v_cache 0.00000560 0.00185397 + layer.2.k_cache 0.00244946 0.46405363 + layer.2.v_cache 0.00001685 0.00453449 + layer.3.k_cache 0.01581429 2.18030964 + layer.3.v_cache 0.00001841 0.00575083 + layer.4.k_cache 0.00072023 0.11370867 + layer.4.v_cache 0.00004652 0.01036312 + layer.4.output 0.15451367 604.12672484 + ------------------------------------------------------------------------------------- + TOTAL 0.07932992 250.77568716 + (elements=765,952) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 765952 +Total Bytes 160000 +BPFP 1.6711 bits/point +EBPFP 3.3422 equivalent bits/point +MSE 250.775687 +---------------------- -------------------------------------------------------- +Time: 0.555s Load: 0.004s, Pack+Encode: 0.227s, Decode+Unpack: 0.324s +---------------------- -------------------------------------------------------- +💾 Converting with 250.7757 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-75.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-75.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-77.zst (97/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-77.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 74, 128) +Output shape: (1, 74, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) -> torch.Size([1, 1, 74, 512]) + layer.4.output: torch.Size([1, 74, 3584]) -> torch.Size([1, 1, 74, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,952B, BPFP=1.0456 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 15,040B, BPFP=3.1757 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,324B, BPFP=1.7576 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 14,272B, BPFP=3.0135 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,224B, BPFP=1.9476 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,552B, BPFP=2.8615 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,712B, BPFP=1.8395 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 14,176B, BPFP=2.9932 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 12,160B, BPFP=2.5676 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,844B, BPFP=2.9231 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 28,256B, BPFP=0.8523 +⌛️ [2/4] FRONTEND: Frontend time: 0.256s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.323s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 74, 128]) + layer.0.v_cache: torch.Size([1, 4, 74, 128]) + layer.1.k_cache: torch.Size([1, 4, 74, 128]) + layer.1.v_cache: torch.Size([1, 4, 74, 128]) + layer.2.k_cache: torch.Size([1, 4, 74, 128]) + layer.2.v_cache: torch.Size([1, 4, 74, 128]) + layer.3.k_cache: torch.Size([1, 4, 74, 128]) + layer.3.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.k_cache: torch.Size([1, 4, 74, 128]) + layer.4.v_cache: torch.Size([1, 4, 74, 128]) + layer.4.output: torch.Size([1, 74, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.10237078 30.90150575 + layer.0.v_cache 0.00001391 0.00484768 + layer.1.k_cache 0.05953315 3.79894195 + layer.1.v_cache 0.00000517 0.00196783 + layer.2.k_cache 0.00728044 0.51170076 + layer.2.v_cache 0.00001648 0.00464016 + layer.3.k_cache 0.03101032 2.52746623 + layer.3.v_cache 0.00001666 0.00558347 + layer.4.k_cache 0.00072555 0.11263505 + layer.4.v_cache 0.00004870 0.01144661 + layer.4.output 0.18364459 725.33554537 + ------------------------------------------------------------------------------------- + TOTAL 0.08744314 300.89585606 + (elements=644,096) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 644096 +Total Bytes 142512 +BPFP 1.7701 bits/point +EBPFP 3.5401 equivalent bits/point +MSE 300.895856 +---------------------- -------------------------------------------------------- +Time: 0.583s Load: 0.004s, Pack+Encode: 0.256s, Decode+Unpack: 0.323s +---------------------- -------------------------------------------------------- +💾 Converting with 300.8959 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-77.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-77.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-8.zst (98/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-8.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.006s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 111, 128) +Output shape: (1, 111, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.0.v_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.1.k_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.1.v_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.2.k_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.2.v_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.3.k_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.3.v_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.4.k_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.4.v_cache: torch.Size([1, 4, 111, 128]) -> torch.Size([1, 1, 111, 512]) + layer.4.output: torch.Size([1, 111, 3584]) -> torch.Size([1, 1, 111, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 6,044B, BPFP=0.8508 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,624B, BPFP=2.7624 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,940B, BPFP=1.5400 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,956B, BPFP=2.6684 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 12,224B, BPFP=1.7207 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,496B, BPFP=2.6036 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,672B, BPFP=1.6430 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,876B, BPFP=2.6571 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 16,276B, BPFP=2.2911 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 18,296B, BPFP=2.5755 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 40,796B, BPFP=0.8204 +⌛️ [2/4] FRONTEND: Frontend time: 0.269s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 111, 128]) + layer.0.v_cache: torch.Size([1, 4, 111, 128]) + layer.1.k_cache: torch.Size([1, 4, 111, 128]) + layer.1.v_cache: torch.Size([1, 4, 111, 128]) + layer.2.k_cache: torch.Size([1, 4, 111, 128]) + layer.2.v_cache: torch.Size([1, 4, 111, 128]) + layer.3.k_cache: torch.Size([1, 4, 111, 128]) + layer.3.v_cache: torch.Size([1, 4, 111, 128]) + layer.4.k_cache: torch.Size([1, 4, 111, 128]) + layer.4.v_cache: torch.Size([1, 4, 111, 128]) + layer.4.output: torch.Size([1, 111, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.311s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 111, 128]) + layer.0.v_cache: torch.Size([1, 4, 111, 128]) + layer.1.k_cache: torch.Size([1, 4, 111, 128]) + layer.1.v_cache: torch.Size([1, 4, 111, 128]) + layer.2.k_cache: torch.Size([1, 4, 111, 128]) + layer.2.v_cache: torch.Size([1, 4, 111, 128]) + layer.3.k_cache: torch.Size([1, 4, 111, 128]) + layer.3.v_cache: torch.Size([1, 4, 111, 128]) + layer.4.k_cache: torch.Size([1, 4, 111, 128]) + layer.4.v_cache: torch.Size([1, 4, 111, 128]) + layer.4.output: torch.Size([1, 111, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.11107565 26.84069274 + layer.0.v_cache 0.00001597 0.00475222 + layer.1.k_cache 0.04333284 3.55649656 + layer.1.v_cache 0.00000558 0.00178348 + layer.2.k_cache 0.00477623 0.43649670 + layer.2.v_cache 0.00001758 0.00463307 + layer.3.k_cache 0.01020195 2.26753936 + layer.3.v_cache 0.00001870 0.00506721 + layer.4.k_cache 0.00079488 0.10736254 + layer.4.v_cache 0.00005041 0.01096804 + layer.4.output 10.27057757 472.01516248 + ------------------------------------------------------------------------------------- + TOTAL 4.23907840 196.31423114 + (elements=966,144) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 966144 +Total Bytes 192200 +BPFP 1.5915 bits/point +EBPFP 3.1830 equivalent bits/point +MSE 196.314231 +---------------------- -------------------------------------------------------- +Time: 0.586s Load: 0.006s, Pack+Encode: 0.269s, Decode+Unpack: 0.311s +---------------------- -------------------------------------------------------- +💾 Converting with 196.3142 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-8.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-8.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-90.zst (99/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-90.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.004s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 71, 128) +Output shape: (1, 71, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) -> torch.Size([1, 1, 71, 512]) + layer.4.output: torch.Size([1, 71, 3584]) -> torch.Size([1, 1, 71, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 4,716B, BPFP=1.0379 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 14,768B, BPFP=3.2500 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 8,180B, BPFP=1.8002 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 13,608B, BPFP=2.9947 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 9,272B, BPFP=2.0405 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 13,136B, BPFP=2.8908 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 8,560B, BPFP=1.8838 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 13,608B, BPFP=2.9947 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 11,892B, BPFP=2.6171 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 13,312B, BPFP=2.9296 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 32,040B, BPFP=1.0073 +⌛️ [2/4] FRONTEND: Frontend time: 0.269s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.280s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 71, 128]) + layer.0.v_cache: torch.Size([1, 4, 71, 128]) + layer.1.k_cache: torch.Size([1, 4, 71, 128]) + layer.1.v_cache: torch.Size([1, 4, 71, 128]) + layer.2.k_cache: torch.Size([1, 4, 71, 128]) + layer.2.v_cache: torch.Size([1, 4, 71, 128]) + layer.3.k_cache: torch.Size([1, 4, 71, 128]) + layer.3.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.k_cache: torch.Size([1, 4, 71, 128]) + layer.4.v_cache: torch.Size([1, 4, 71, 128]) + layer.4.output: torch.Size([1, 71, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.09375787 30.98723592 + layer.0.v_cache 0.00001622 0.00485382 + layer.1.k_cache 0.05984993 3.87452418 + layer.1.v_cache 0.00000512 0.00175502 + layer.2.k_cache 0.00243208 0.47125776 + layer.2.v_cache 0.00001865 0.00530229 + layer.3.k_cache 0.04944374 2.14599266 + layer.3.v_cache 0.00001908 0.00572066 + layer.4.k_cache 0.00084877 0.11752403 + layer.4.v_cache 0.00005139 0.01174003 + layer.4.output 0.19371969 749.60928068 + ------------------------------------------------------------------------------------- + TOTAL 0.09191063 310.87593360 + (elements=617,984) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 617984 +Total Bytes 143092 +BPFP 1.8524 bits/point +EBPFP 3.7047 equivalent bits/point +MSE 310.875934 +---------------------- -------------------------------------------------------- +Time: 0.554s Load: 0.004s, Pack+Encode: 0.269s, Decode+Unpack: 0.280s +---------------------- -------------------------------------------------------- +💾 Converting with 310.8759 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-90.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-90.zst + + 💪 Processing: ../datasets/KimiAudio-7B-Instruct-500features-L5wCache/KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-98.zst (100/100) + +[1/4] FRONTEND: Loading features from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-98.zst... + +Original data structure: +⌛️ [1/4] FRONTEND: Load time: 0.005s + +------------------------------------------------------------ +KimiAudio Features Summary +------------------------------------------------------------ +Number of layers: 5 +Layer indices: [0, 1, 2, 3, 4] +Last layer index: 4 +Cache shape: (1, 4, 109, 128) +Output shape: (1, 109, 3584) +Data type: torch.bfloat16 +Has output output: True +------------------------------------------------------------ + +[2/4] FRONTEND: Pack + Encode (strategy: individual)... + IndividualPacker: + layer.0.k_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.0.v_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.1.k_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.1.v_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.2.k_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.2.v_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.3.k_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.3.v_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.4.k_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.4.v_cache: torch.Size([1, 4, 109, 128]) -> torch.Size([1, 1, 109, 512]) + layer.4.output: torch.Size([1, 109, 3584]) -> torch.Size([1, 1, 109, 3584]) + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + layer.0.k_cache: 5,936B, BPFP=0.8509 + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + layer.0.v_cache: 19,104B, BPFP=2.7385 + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + layer.1.k_cache: 10,712B, BPFP=1.5356 + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + layer.1.v_cache: 18,548B, BPFP=2.6588 + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + layer.2.k_cache: 11,952B, BPFP=1.7133 + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + layer.2.v_cache: 18,040B, BPFP=2.5860 + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + layer.3.k_cache: 11,264B, BPFP=1.6147 + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + layer.3.v_cache: 18,516B, BPFP=2.6542 + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + layer.4.k_cache: 15,912B, BPFP=2.2810 + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + layer.4.v_cache: 17,836B, BPFP=2.5568 + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + layer.4.output: 42,044B, BPFP=0.8610 +⌛️ [2/4] FRONTEND: Frontend time: 0.249s (Pack+Encode) + +[3/4] BACKEND: Decode + Unpack... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 109, 128]) + layer.0.v_cache: torch.Size([1, 4, 109, 128]) + layer.1.k_cache: torch.Size([1, 4, 109, 128]) + layer.1.v_cache: torch.Size([1, 4, 109, 128]) + layer.2.k_cache: torch.Size([1, 4, 109, 128]) + layer.2.v_cache: torch.Size([1, 4, 109, 128]) + layer.3.k_cache: torch.Size([1, 4, 109, 128]) + layer.3.v_cache: torch.Size([1, 4, 109, 128]) + layer.4.k_cache: torch.Size([1, 4, 109, 128]) + layer.4.v_cache: torch.Size([1, 4, 109, 128]) + layer.4.output: torch.Size([1, 109, 3584]) +⌛️ [3/4] BACKEND: Backend time: 0.297s + +[4/4] METRICS: Computing MSE Breakdown... + Using per-key quantization points (layer.0.k_cache: torch.Size([256])) for layer.0.k_cache + Using per-key quantization points (layer.0.v_cache: torch.Size([256])) for layer.0.v_cache + Using per-key quantization points (layer.1.k_cache: torch.Size([256])) for layer.1.k_cache + Using per-key quantization points (layer.1.v_cache: torch.Size([256])) for layer.1.v_cache + Using per-key quantization points (layer.2.k_cache: torch.Size([256])) for layer.2.k_cache + Using per-key quantization points (layer.2.v_cache: torch.Size([256])) for layer.2.v_cache + Using per-key quantization points (layer.3.k_cache: torch.Size([256])) for layer.3.k_cache + Using per-key quantization points (layer.3.v_cache: torch.Size([256])) for layer.3.v_cache + Using per-key quantization points (layer.4.k_cache: torch.Size([256])) for layer.4.k_cache + Using per-key quantization points (layer.4.v_cache: torch.Size([256])) for layer.4.v_cache + Using per-key quantization points (output: torch.Size([256])) for layer.4.output + IndividualUnPacker: + layer.0.k_cache: torch.Size([1, 4, 109, 128]) + layer.0.v_cache: torch.Size([1, 4, 109, 128]) + layer.1.k_cache: torch.Size([1, 4, 109, 128]) + layer.1.v_cache: torch.Size([1, 4, 109, 128]) + layer.2.k_cache: torch.Size([1, 4, 109, 128]) + layer.2.v_cache: torch.Size([1, 4, 109, 128]) + layer.3.k_cache: torch.Size([1, 4, 109, 128]) + layer.3.v_cache: torch.Size([1, 4, 109, 128]) + layer.4.k_cache: torch.Size([1, 4, 109, 128]) + layer.4.v_cache: torch.Size([1, 4, 109, 128]) + layer.4.output: torch.Size([1, 109, 3584]) + Per-key MSE Breakdown (Quant=Truncation+Quantization, Codec=Compression): + Key Quant-MSE Total-MSE + ------------------------------------------------------------------------------------- + layer.0.k_cache 0.12860953 29.00083097 + layer.0.v_cache 0.00001500 0.00500611 + layer.1.k_cache 0.05917679 3.40419328 + layer.1.v_cache 0.00000528 0.00197846 + layer.2.k_cache 0.01253135 0.49999265 + layer.2.v_cache 0.00001869 0.00491830 + layer.3.k_cache 0.07406004 2.29168197 + layer.3.v_cache 0.00001854 0.00573805 + layer.4.k_cache 0.00071234 0.11368437 + layer.4.v_cache 0.00004916 0.01128456 + layer.4.output 10.45909253 488.90010649 + ------------------------------------------------------------------------------------- + TOTAL 4.32287320 203.39059142 + (elements=948,736) +---------------------- -------------------------------------------------------- +SAMPLE-WISE STATISTICS +---------------------- -------------------------------------------------------- +Handler kimiaudio +Strategy individual +Architecture hyperprior-featurecoding +---------------------- -------------------------------------------------------- +Total Elements 948736 +Total Bytes 189864 +BPFP 1.6010 bits/point +EBPFP 3.2020 equivalent bits/point +MSE 203.390591 +---------------------- -------------------------------------------------------- +Time: 0.551s Load: 0.005s, Pack+Encode: 0.249s, Decode+Unpack: 0.297s +---------------------- -------------------------------------------------------- +💾 Converting with 203.3906 MSE: + from ../datasets/KimiAudio-7B-Instruct-500features-L5wCache//KimiAudio-7B-Instruct-500features-L5wCache/sd-qa/SD-QA-sd-qa-98.zst + to output-fixed/kimiaudio/lambda0.02/hyperprior-featurecoding-8bit-individual/sd-qa/SD-QA-sd-qa-98.zst +------------------------ ---------------------------- +TOTAL PROCESSING SUMMARY +------------------------ ---------------------------- +Total files 100 +Avg BPFP 1.7714 bits/point +Avg EBPFP 3.5427 equivalent bits/point +Avg MSE 307.990394 +Avg Time 0.498s +------------------------ ----------------------------