Spaces:

HOLOGRAMTECH
/

q

Running

App Files Files Community

q

7.14 MB

Ctrl+K

Ctrl+K

1 contributor

History: 46 commits

Humuhumu33's picture

warmup uses decode() to precompile batched pipelines (cut first-msg TTFT)

e4b6a8f verified 9 minutes ago

core
decode(): report live-path GPU ms/tok for clean CPU/GPU split about 1 hour ago
forge
Upload folder using huggingface_hub about 14 hours ago
pkg
Upload folder using huggingface_hub about 14 hours ago
.gitattributes

1.52 kB
initial commit about 14 hours ago
README.md

184 Bytes
initial commit about 14 hours ago
blake3-gpu-parallel.html

13.4 kB
Upload blake3-gpu-parallel.html with huggingface_hub about 11 hours ago
blake3-gpu-test.html

15.3 kB
Upload blake3-gpu-test.html with huggingface_hub about 11 hours ago
gemv-bench.html

11.7 kB
ternary GEMV kernel lab: V0/V1/V2 variants about 4 hours ago
gpu-blake3.mjs

9 kB
Upload gpu-blake3.mjs with huggingface_hub about 5 hours ago
holo-blake3.mjs

7.44 kB
Upload holo-blake3.mjs with huggingface_hub about 11 hours ago
holo-delta.mjs

10.8 kB
Upload folder using huggingface_hub about 14 hours ago
holo-gpu-device.mjs

1.45 kB
Upload holo-gpu-device.mjs with huggingface_hub about 5 hours ago
holo-load-delta.mjs

3.81 kB
Upload folder using huggingface_hub about 14 hours ago
holo-load2bit.mjs

12.2 kB
Upload holo-load2bit.mjs with huggingface_hub about 5 hours ago
holo-model-frame.mjs

6.48 kB
Upload folder using huggingface_hub about 14 hours ago
index.html

33.2 kB
warmup uses decode() to precompile batched pipelines (cut first-msg TTFT) 9 minutes ago
kernel-profile.html

11.5 kB
profiler: fix read-only binding + dot8 batched-verify proxy + refined verdict about 3 hours ago
q-chat.html

9 kB
Upload folder using huggingface_hub about 14 hours ago
qvac-2bit.mjs

20 kB
Upload folder using huggingface_hub about 14 hours ago
qvac-gpu.js

143 kB
decode(): report live-path GPU ms/tok for clean CPU/GPU split about 1 hour ago
qvac-ingest.mjs

28.6 kB
Upload folder using huggingface_hub about 14 hours ago
rb-bench.html

9.92 kB
register-blocked GEMV lab: B=1/2/4/8/16 rows per group about 3 hours ago
roofline.html

8.91 kB
roofline: batched passes + onSubmittedWorkDone for real bw about 4 hours ago
style.css

388 Bytes
initial commit about 14 hours ago