rustytorch

326 Commits 1 Branch 0 Tags

3 Commits

Include renames

Author	SHA1	Message	Date
osobhandClaude Opus 4.7	a5cedfb46a	rtx-csm: emotional_speech_guide — CREMA-D vs RAVDESS firdhokk verdict 8-gen bench (4 emotions × 2 corpora) at seed=42 against firdhokk Whisper-LV3: target RAVDESS CREMA-D happy happy (0.999) ✓ happy (0.999) ✓ angry neutral (0.92) sad (0.99) fearful happy (0.998) fearful (0.984) ✓ sad angry (0.99) fearful (0.99) CREMA-D 2/4 vs RAVDESS 1/4. Larger / more naturalistic corpus produces more class-pure fearful direction. Neither corpus solves angry or sad — recipe shifts into 'vague expressivity' rather than class-specific corners. Practical: prefer CREMA-D when available; A/B both per emotion if class precision matters. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>	2026-04-30 00:01:02 -07:00
osobhandClaude Opus 4.7	479cd69bc3	rtx-csm: WavLM-SV parity scaffolding for Phase 5d - scripts/wavlm_sv_parity.py: Python-side reference embedder. Loads HF WavLMForXVector + Wav2Vec2FeatureExtractor and dumps a JSON fingerprint (cosine + per-utterance norm + first/last 8 elements) for comparison. - examples/wavlm_sv_demo gains --parity-json flag emitting the same fingerprint structure on the Rust side. Once the user has a Python env with transformers + torch installed, running both produces side-by-side JSON files for diffing — first-pass sanity check on whether our port matches HF numerically. We can't run the Python side from this Rust shell. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>	2026-04-25 21:18:53 -07:00
osobhandClaude Opus 4.7	d1dba7a05c	rtx-csm: WavLM-SV converter + end-to-end speaker similarity Phase 5c — pure-Rust converter for microsoft/wavlm-base-plus-sv with auto-detecting weight_norm merger; verified on real 100M-param weights: load + embed + cosine-similarity round-trip works on Metal. - wavlm_sv_convert.rs: candle_core::pickle reads pytorch_model.bin directly. merge_weight_norm_auto picks the kept dim from g's shape (dim=0 for AudioSeal SEANet, dim=2 for WavLM pos_conv_embed). Skips classifier./objective. (train-only AMSoftmax head). - examples/wavlm_sv_convert: HF download + convert CLI. Verified output: 1 weight_norm pair merged + 261 passthrough + 3 skipped = 262 tensors. - examples/wavlm_sv_demo: load + embed pair of WAVs + cosine similarity. - examples/audioseal_inspect: gains --which wavlm-sv variant for key discovery. - hub.rs: REPO_WAVLM_SV + resolve_wavlm_sv() helper. - wavlm_sv::XVectorHead bug fix: layer_weights is top-level, not nested under prefix. Verified end-to-end on Metal: cosine sim 0.9985 on same-speaker pair (CSM vs CSM-watermarked, 10s @ 24 kHz resampled to 16 kHz). Numerical parity vs HF reference is Phase 5d. 3 converter tests + 12 wavlm_sv tests; 78 lib tests total green. Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>	2026-04-25 19:57:52 -07:00

Author

SHA1

Message

Date

osobhandClaude Opus 4.7

a5cedfb46a

rtx-csm: emotional_speech_guide — CREMA-D vs RAVDESS firdhokk verdict

8-gen bench (4 emotions × 2 corpora) at seed=42 against firdhokk
Whisper-LV3:

  target    RAVDESS              CREMA-D
  happy     happy (0.999) ✓      happy (0.999) ✓
  angry     neutral (0.92)       sad (0.99)
  fearful   happy (0.998)        fearful (0.984) ✓
  sad       angry (0.99)         fearful (0.99)

CREMA-D 2/4 vs RAVDESS 1/4. Larger / more naturalistic corpus
produces more class-pure fearful direction. Neither corpus solves
angry or sad — recipe shifts into 'vague expressivity' rather than
class-specific corners.

Practical: prefer CREMA-D when available; A/B both per emotion if
class precision matters.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

2026-04-30 00:01:02 -07:00

osobhandClaude Opus 4.7

479cd69bc3

rtx-csm: WavLM-SV parity scaffolding for Phase 5d

- scripts/wavlm_sv_parity.py: Python-side reference embedder. Loads HF
  WavLMForXVector + Wav2Vec2FeatureExtractor and dumps a JSON fingerprint
  (cosine + per-utterance norm + first/last 8 elements) for comparison.
- examples/wavlm_sv_demo gains --parity-json flag emitting the same
  fingerprint structure on the Rust side.

Once the user has a Python env with transformers + torch installed,
running both produces side-by-side JSON files for diffing — first-pass
sanity check on whether our port matches HF numerically. We can't run
the Python side from this Rust shell.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

2026-04-25 21:18:53 -07:00

osobhandClaude Opus 4.7

d1dba7a05c

rtx-csm: WavLM-SV converter + end-to-end speaker similarity

Phase 5c — pure-Rust converter for microsoft/wavlm-base-plus-sv with
auto-detecting weight_norm merger; verified on real 100M-param weights:
load + embed + cosine-similarity round-trip works on Metal.

- wavlm_sv_convert.rs: candle_core::pickle reads pytorch_model.bin
  directly. merge_weight_norm_auto picks the kept dim from g's shape
  (dim=0 for AudioSeal SEANet, dim=2 for WavLM pos_conv_embed). Skips
  classifier.*/objective.* (train-only AMSoftmax head).
- examples/wavlm_sv_convert: HF download + convert CLI. Verified output:
  1 weight_norm pair merged + 261 passthrough + 3 skipped = 262 tensors.
- examples/wavlm_sv_demo: load + embed pair of WAVs + cosine similarity.
- examples/audioseal_inspect: gains --which wavlm-sv variant for key
  discovery.
- hub.rs: REPO_WAVLM_SV + resolve_wavlm_sv() helper.
- wavlm_sv::XVectorHead bug fix: layer_weights is top-level, not nested
  under prefix.

Verified end-to-end on Metal: cosine sim 0.9985 on same-speaker pair
(CSM vs CSM-watermarked, 10s @ 24 kHz resampled to 16 kHz). Numerical
parity vs HF reference is Phase 5d.

3 converter tests + 12 wavlm_sv tests; 78 lib tests total green.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

2026-04-25 19:57:52 -07:00