Hugging Face's logo

stillerman
/
fdt-disfluency-tiny-4m

Token Classification
Transformers.js
ONNX
Safetensors
bert
disfluency-detection
Model card Files
xet
 Community
fdt-disfluency-tiny-4m / README.md
stillerman's picture
stillerman
publish disfluency tagger
ba012f0 verified
preview code
|
Raw
History Blame Contribute Delete
1.11 kB
metadata 
license: apache-2.0
pipeline_tag: token-classification
tags:
  - disfluency-detection
  - transformers.js
  - onnx

stillerman/fdt-disfluency-tiny-4m

Disfluency deletion tagger for live speech transcripts: tags every whitespace word KEEP / DELETE / KEEP_STRIP_COMMA / KEEP_CAPITALIZE, then a ~15-line reconstruction turns tags into cleaned text. Deletion-only by construction — it cannot rephrase, hallucinate, or alter names and numbers.

  • Architecture: BERT L2/H128 (4.4M params), v1 data mix
  • Val metrics: exact-match 0.5980, DELETE-F1 0.8892
  • Training data: synthetic disfluency injection over conversational corpora — see stillerman/fdt-disfluency-synthetic
  • onnx/model_quantized.onnx (int8) is ready for transformers.js (device: "webgpu", dtype: "q8"); runs at ~10–50 ms per utterance in-browser.
  • ⚠️ Trained partly on DailyDialog (CC BY-NC-SA): treat as research artifact, not for commercial deployment as-is.

Trained on a DGX Spark as part of the FluencyAI digital-twin project.