Hugging Face's logo

mazesmazes
/
tiny-audio

Automatic Speech Recognition
Transformers
Safetensors
English
asr_model
feature-extraction
asr
speech-recognition
audio
qwen
glm-asr
custom_code
Model card Files
xet
 Community
mazesmazes commited on
Commit
206c181
·
verified ·
1 Parent(s): f3bab70

Revert to 5a38762f (end of stage 1) — drop bad stage 2 checkpoints

Browse files

Stage 2 (started 2026-05-09 16:16) regressed across most evals due to over-aggressive augmentations (MP3 at 32 kbps + babble + bandlimit compounding). This commit restores file contents to the end-of-stage-1 checkpoint (f3bab709 → 5a38762f). Stage 2 history is preserved above.

Files changed (2) hide show
  1. model.safetensors +1 -1
  2. training_args.bin +1 -1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:113b7f29d9cf01831d135ea725d406a47963391cdfcf7e363030b915f0a6dc9c
3
  size 2433494416
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e830a9e4362b32186ad529b33d2fb7c7c930ca182d471178f2d24bee05e3bb4
3
  size 2433494416
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2ac9e08dd8f7aca66959a98df29ac982b3d174e01b41ca39a0e2535d2af5b6ce
3
  size 5329
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:323f3fcb8fda7d037add78c657f2f95394b69825fe27e5f794b4b2607076e45a
3
  size 5329