llama.cpp

Files

T

Stephen CoxandGitHub 547765a93e mtmd: add Gemma 4 audio conformer encoder support (#21421 )

* mtmd: add Gemma 4 audio conformer encoder support

Add audio processing for Gemma 4 E2B/E4B via a USM-style Conformer.

Architecture:
- 12-layer Conformer: FFN → Self-Attention → Causal Conv1D → FFN → Norm
- Subsampling Conv Projection: 2x Conv2D(stride=2) with LayerNorm
- Full self-attention with sinusoidal RPE and sliding window mask (24)
- Logit softcapping at 50.0, ClippableLinear clamping
- Output: 1024 → 1536 → RMSNorm → multimodal embedder

Mel preprocessing (dedicated mtmd_audio_preprocessor_gemma4a):
- HTK mel scale, 128 bins, magnitude STFT, mel_floor=1e-3
- Standard periodic Hann window (320 samples), zero-padded to FFT size
- Semicausal left-padding (frame_length/2 samples)
- Frame count matched to PyTorch (unfold formula)
- No pre-emphasis, no Whisper-style normalization
- Mel cosine similarity vs PyTorch: 0.9998

Key fixes:
- Tensor loading dedup: prevent get_tensor() from creating duplicate
  entries in ctx_data. Fixed with std::set guard.
- ClippableLinear clamp_info loading moved after per-layer tensors.
- Sliding window mask (24 positions) matching PyTorch context_size.
- Skip Whisper normalization for Gemma4 mel output.

Tested on E2B and E4B with CPU and Vulkan backends.
Transcribes: "Glad to see things are going well and business is starting
to pick up" (matching ground truth).

Ref: #21325

2026-04-12 14:15:26 +02:00

cogvlm.cpp

mtmd: add clip_graph::build_mm() (#20751 )

2026-03-19 13:11:39 +01:00

conformer.cpp

mtmd: add clip_graph::build_mm() (#20751 )

2026-03-19 13:11:39 +01:00

deepseekocr.cpp

mtmd: Add DeepSeekOCR Support (#17400 )

2026-03-25 19:57:40 +01:00

dotsocr.cpp

mtmd: support dots.ocr (#17575 )