Add files using upload-large-folder tool

README.md

@@ -19,9 +19,10 @@ size_categories:
 
 # DramaBox reinterpretations — reward-ranked top 3 of 64
 
-**~10,000 acting prompts, each re-performed 64 times by
+**~20,000 acting prompts, each re-performed 64 times by
 [`laion/moss-tts-local-transformer-4.55b-voice-acting-v2`](https://huggingface.co/laion/moss-tts-local-transformer-4.55b-voice-acting-v2),
-with the three highest-reward takes published here.** Roughly 30,000 clips.
+with the three highest-reward takes published here — plus the original DramaBox clip each one
+reinterprets.** Roughly 80,000 audio files (~20k originals + ~60k takes).
 
 The companion release [`…-raw64`](#) keeps all 64 candidates per group, so the full reward
 distribution — not just its upper tail — stays available.
@@ -45,15 +46,19 @@ Median WER against the intended script is **0.17**, and only **1.6 %** of candid
 exactly. That is largely the disfluent source text, not model failure — the acoustic quality is
 much better than the WER suggests.
 
-## Scope — please read the number carefully
+## Scope
 
-This release is the **edge-case half**: **10,000 groups / 30,009 clips**, spread over **262
-distinct edge-case categories** (`ec000` … `ec261`), roughly evenly.
+**Complete: ~20,000 groups over both halves.**
 
-The **second half** — 9,980 groups over the non-edge DramaBox material (`prod1m`, the
-Gemma-authored acting prompts, stratified by `pathway`) — is generating now and will be added to
-this same repository, bringing the corpus to **~20,000 groups**. Until that lands, cite the
-number you actually see in `manifest.parquet` rather than 20k.
+- the **edge-case half** — 10,000 groups over **262 distinct edge-case categories**
+  (`ec000` … `ec261`), roughly evenly;
+- the **prod1M half** — ~10,000 groups over the non-edge DramaBox material (the Gemma-authored
+  acting prompts, stratified by `pathway`).
+
+Each group now ships **the original DramaBox clip alongside its three reinterpretations**, so a
+take can be compared against what it reinterprets without resolving `src_id` anywhere else.
+
+Cite the number in `manifest.parquet` rather than a round figure.
 
 ## Contents
 
@@ -63,7 +68,11 @@ number you actually see in `manifest.parquet` rather than 20k.
 | `data/top3-NNNN.parquet` | per-shard manifest (key, reward, WER, duration, blend, genuineness, burst count) |
 | `manifest.parquet` | all shards concatenated |
 
-Key format: `<src_id>.top<rank>`, rank ∈ {1,2,3} by descending reward.
+Key format: `<src_id>.top<rank>`, rank ∈ {1,2,3} by descending reward, plus **`<src_id>.orig`**
+— the original DramaBox source clip for that group, so every group is four audio files. The
+original's JSON carries `is_original: true`, `rank: 0` and the prompt, but no reward or model
+scores (it is not a candidate). Filter it out with `is_original` or `rank > 0` when you want only
+the generations.
 
 ### Per-clip JSON
 
@@ -75,8 +84,8 @@ regression value, bucket and natural-language label), `quality` (4 dims), `burst
 `prompt_general`, `prompt_script`) and the source identity (`src_id`, `src_duration`,
 `category`).
 
-**Source audio is not duplicated here** — `src_id` resolves against the DramaBox staging
-corpus.
+**The original source audio IS included** as `<src_id>.orig.mp3` (see key format above), so no
+lookup against the DramaBox staging corpus is needed.
 
 ## The LoRAs used
 
@@ -157,7 +166,33 @@ Over a 38,400-candidate sample of the full pool (i.e. before top-3 selection):
 - Top-3 selection means this release is the **upper tail** of the reward distribution by
   construction. For unbiased analysis use the raw-64 companion.
 - The `top3` flag was verified against a fresh ranking by `reward_v1c` on 400 groups —
-  **400/400 agreement**.
+  **400/400 agreement**. Note that this check sampled the **edge half only**, which is exactly
+  why it did not catch the issue below.
+
+### A ranking bug that affected the prod1M half (fixed before release)
+
+Worth stating plainly, because it is the kind of failure that ships silently.
+
+The reward's similarity term `s` compares a candidate against the **source clip's** emotion
+profile, so it needs the source clip scored. The edge half carries those scores already; the
+prod1M half does not, and the step that computes them
+(`src_scores_prod1m.parquet`) **was never run**. The annotator's reward block is guarded by a
+lookup of that table, so for every prod1M group it was skipped, leaving each candidate without a
+`reward_v1c`. Ranking then fell through to `sorted(..., key=lambda i: -c.get("reward_v1c", -1e9))`
+— every candidate tied, Python's sort is stable, and **the input order survived as the
+"ranking"**. The "top 3" for that half was candidates 0, 1, 2 of 64.
+
+Nothing downstream could tell: three candidates were flagged per group, `rank_in_group` was
+populated 0…63, and the schema validated. Only the *values* gave it away — a real ranking is a
+permutation, and this one was the identity.
+
+**Resolution.** The source vectors were computed, `reward_v1c` was recomputed for every affected
+group, and the shards were rebuilt. The packer now refuses to run if the source-score table is
+missing, and a gate samples the rescored groups and aborts if `top3` is still `[0,1,2]`. The
+published shards are ranked by the reward described above throughout.
+
+**If you downloaded this dataset before 2026-08-08**, the prod1M shards you have are arbitrary
+picks rather than top-3. The edge-half shards were always correctly ranked.
 
 ## License
 

data/top3-0000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8591ac52ac633c96f2a4cc1dc7e8bb4f52463af3d235607651d42f4f43a76a00
-size 177455
+oid sha256:748e356e8f8baa7fe3a17110f36c095113696e475914f17c55eed42a41cb8300
+size 193624

data/top3-0000.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f773ffff5fbea523df9058a91797367c17d29f4b9f4f195383903921daeb70d0
-size 618024960
+oid sha256:8cd6b146cbc68f894df6e685d495b1b083a652c336e64d4afd945c9585c22654
+size 898263040

data/top3-0001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e9e1e9edec3ccaccb5a682172fca61efb5aaadcbbf9e5ed562040dc3d0c0d15d
-size 178355
+oid sha256:dba97331dd5e8df6c6de5718e3bad1756bdeb01e2f5f1fd3fca131b02ea8dbbb
+size 194485

data/top3-0001.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae61c66b7d59bacea51bda7d1f1ca003d153921878bdd70d7990f0fd44de265c
-size 624844800
+oid sha256:07755acf779871492c3d154b360a2bfa1a058cfdbc717a492f16c9541ef63c26
+size 906915840

data/top3-0002.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:933b731a4ef068416e5d52f6b591ebc34c1cb63ce01cabec90ea8d5b76f29d9a
-size 178134
+oid sha256:8668f989d82ee9993fa41f8b357cf6ca11ff762ea164151168719547f27094a4
+size 194278

data/top3-0002.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfaf02c1165d8a47c6808c837bf9e9377e8da249988caa01f6668712b1921dc9
-size 625571840
+oid sha256:2eae465c0c6c421c0c5df5cfb66195959a0c89ece633bf2b2e87d4d5010f1a65
+size 909015040

data/top3-0003.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:edd0c69ff3b764d06feebeacb67e19540f93de2c991f46b4d5ad36ae7779e684
-size 178362
+oid sha256:5e31f0bc93682be88ad146adfb54e600d2dccf511d2795a0994998a262c6e2a3
+size 194638

data/top3-0003.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:827bf06099abb671ea9638125a4bacf24756a7a75301923a202d7421265697eb
-size 630353920
+oid sha256:ad84908d8c30577741e923b92ac96b2d265dfe6a0d4cb4e05564b71c322ac73b
+size 914268160

data/top3-0004.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d655867bd4f68edcba4b26c577e869d14a8b4a65d0b4b2020619a2e4f90d2d2a
-size 177899
+oid sha256:a8ce0c88f36e9a684568be21ed33faeb4ce19931604a0344cf99da1249f65a73
+size 194096

data/top3-0004.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9eaf523f3ede6407bd0721673ffe3a0e82d5146921dace90b3d7cdf4b544846
-size 621168640
+oid sha256:566553ea4e2ef88d07879f6edcb291ed8b617095a6cab6f16fc23d393a029159
+size 902215680

data/top3-0005.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9fac7321a901a3310834210fa81a9d7688456a0b71f0ae61b123ed92b4d0b169
-size 178089
+oid sha256:a4fde10800ddee64b8a26560df8ab6fa1ddf552de41e64dba85d177438959723
+size 194391

data/top3-0005.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2a513f4ce37457add5da86feb54e95ea61688f5a6cbbf8b2af4717094ec106f
-size 617707520
+oid sha256:b705fba7b4758d1a693c122d3487f9ffaa32db8886837d4d355d3f5ec0be899a
+size 898805760

data/top3-0006.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9a53f5f483450c0b659e44e1c7b350b1ace559912f0c77f34d201e70415e0af6
-size 178541
+oid sha256:f8e28616b010d7e2d75fdb8dcbdc2887f446d76cdb473c8a53100bf3e1f1ea46
+size 194771

data/top3-0006.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2477472f50e973c2310506c5f5b38f82e7d992ecd93d6128f2a55a31778c3fd3
-size 621455360
+oid sha256:a30b98657dfc5d8bf9a12bd8e2b66d30c81fdeafc87afbe6ad9b4b1e0b59f4cd
+size 903925760

data/top3-0007.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8910a98c137b1029050f4b876b34f677711a9cb81c7f38df8d39702e0908a8ea
-size 178377
+oid sha256:d851952f787fbe1d6f7ebfde0779f56720049a5ee9d2ca0c1f2139800e9063ea
+size 194498

data/top3-0007.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:85ed2e45af3cc14b727821b2dfd97b6f6cbb3473bc88538cea4e4f9d64f5c55c
-size 615383040
+oid sha256:9b2bae6f000d587e47c9399762bec36b12bcb9c176fd9f4c831dd071d6f86868
+size 896071680

data/top3-0008.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b4a17cbdf9317b776af80fe8d3a7dc91d158a705d06063685c82ffecdc2e687
-size 178003
+oid sha256:92979729a9222002e2182f351f1792465c197e710d162d29a86248e68fdd9894
+size 194184

data/top3-0008.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ca793a85190b003c4276f0660c06e6021cdfebe27637c790c4cae543c436f14f
-size 619847680
+oid sha256:3d8cfdcb193770b462a9fce73fffb776ce2179597d92970ab41ab7e5e56570e9
+size 901550080

data/top3-0009.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:efeac47227cdc9ba87883fa5386f502f271e274c4de3eacbe4fac8dbf06ade64
-size 177749
+oid sha256:d260eb81088589cddc9b60d5907c960b0b6d1bc475207100b24598ceae4cfe30
+size 193803

data/top3-0009.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:60abca9fca2029d2ad1efa2440c294427d1cbc968f0037b8f4369f02e4b9d1ee
-size 614676480
+oid sha256:81965e49e2be1934ab13a24a88d56532c809f8bddafe059b6416f7f2b5509f87
+size 894853120

data/top3-0010.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:50e4220f788be51c9ff08abd4d5892a587f32c02d7f7e0951f32e93fe59065c7
-size 9024
+oid sha256:9d80009b4532e16e221a20902022f45be4b84520dcbb212f61324623e93ed676
+size 231764

data/top3-0010.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e39acfd6ca62bb0f0961445df00cda8af8c1ab4734065c7d1d019a4e6beaacd8
-size 3328000
+oid sha256:f5f2c43e92e26eb31c1034827ff2820d9ab4a10583a34c3bd78220888ca64e7c
+size 858357760

data/top3-0011.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b39afdc5f6b7d89715b3aad821f3c3141babf3087c272cdd01c0f52370ad0629
+size 232791

data/top3-0011.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24936f3442296c179a1e0ad99cca99e0bedc3ed18070ca3518e7a693fc1c0a5b
+size 852766720

data/top3-0012.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57944c31b5433cc158e7128b7afc37404b61b51a4c9b0ce53d80f15037ad9b2d
+size 232081

data/top3-0012.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a91986d52bb349e2410b1927e79128f709c72d50b122e59e1da2a519ac92fe54
+size 848875520

data/top3-0013.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87572328c1b8a584e5976863b2bf0b9c8ee0f16f3b5795bcea32d8ac4a2db2ae
+size 232469

data/top3-0013.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3f72d0b41b488f10a2b989749444d320314f81803d1cada907c6e30dde7ef3f1
+size 852162560

data/top3-0014.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e37b5ee20182c59d5b1432134bd855684d267e3af52e444f77a1c4a8fc2d54c4
+size 233071

data/top3-0014.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9a438073054367b5c0220860bacf270b885ca9ec75b7ae35bf25885a086f1248
+size 855101440

data/top3-0015.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e402bab0c4c179fa10fbdecf5fb161337ae78912515166fa36b3302d2f0dbbfb
+size 233424

data/top3-0015.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7dc703abd159d9c402e6cb470c978783933c5e65edb269a2df40fddd0fccd8b
+size 862105600

data/top3-0016.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb3891bd1f64f4f486cb0e9923a215466e0cbb5c434b01ec6da0eb6f1e2e1a6d
+size 233862

data/top3-0016.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af1a3ce70f5e49f7c958c286785f643ff90a3dc97ec5bf301ea742690eb18a6e
+size 861614080

data/top3-0017.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9c27e698dda32f2a4d25c2ac45463e566cd473ca738b05082b23cf249807771
+size 234455

data/top3-0017.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:223634ac1fa0b4bfc731ac75277ab205446d2a1d26c6bdb20f25dee0bef54d49
+size 862679040

data/top3-0018.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:990ccfafb1d8827a23053cd8abc5156ae1195a382f64372f161252ab00012be0
+size 234187

data/top3-0018.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ec51281eb79103d3e2982f7ce6a993927b70cf2e4e321e04948b63c89c3509e
+size 865300480

data/top3-0019.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a14789ee50242ab6ecde6329d00b6d20efdf5e4b7bca4c1a9d3e996f7fe25930
+size 229932

data/top3-0019.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:287a661ab99149d9fce187dd085962f2ee6c877482a630e64e3a69f486d5228a
+size 839516160

manifest.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c2e44269d72bf69747552dd3e297a7d05ce3ee62e984da971e6b9dcf46fe99a3
-size 1511113
+oid sha256:38479a1d7ff8cc5be9286edfddfd9379dd47c2dc98ddc02a440b4e0d51b20203
+size 3638245