---
library_name: peft
pipeline_tag: automatic-speech-recognition
language:
- bm
- fr
- en
tags:
- lora
- peft
- whisper
- automatic-speech-recognition
- tensorboard
datasets:
- djelia/bambara-asr
- djelia/bambara-synthetic-audio
- djelia/bambara-audio-b
- sudoping01/bambara-speech-recognition-benchmark
---

# bm-whisper-large-v4-training-bm-lora-3c

A decoder-only, rank-8 LoRA adapter for Bambara speech recognition, sized for Whisper large-v3 geometry: hidden size 1280, 32 encoder + 32 decoder layers, 128 mel bins, 51,866-token vocabulary.

Adapter weights only. The base checkpoint is not recorded in this repo, so supply your own Whisper large-v3-geometry model when loading.

## Usage

```python
from peft import PeftModel
from transformers import WhisperForConditionalGeneration, WhisperProcessor

base = WhisperForConditionalGeneration.from_pretrained(YOUR_BASE_MODEL)
model = PeftModel.from_pretrained(base, "djelia/bm-whisper-large-v4-training-bm-lora-3c")
model.eval()

# tokenizer and feature extractor ship with the adapter
processor = WhisperProcessor.from_pretrained("djelia/bm-whisper-large-v4-training-bm-lora-3c")

# Optional: fold the LoRA deltas into the base weights for inference.
# merged = model.merge_and_unload()
```

## Adapter configuration

| Key | Value |
|---|---|
| `peft_type` | `LORA` |
| `r` / `lora_alpha` | 8 / 8 (scaling 1.0) |
| `lora_dropout` | 0.05 |
| `bias` / `lora_bias` | `none` / false |
| `target_modules` | regex `model.decoder.layers.[\d]+.(self_attn\|encoder_attn).(q_proj\|k_proj\|v_proj\|out_proj)` |
| Trainable parameters | 5,242,880 |
| Adapter dtype | F32 |

## Notes

`target_modules` is a regex scoped to `model.decoder.layers`, so the adapter covers decoder self-attention and cross-attention `q/k/v/out_proj` across all 32 decoder layers (512 tensors) and nothing else. The encoder, every MLP block, all embeddings and all layer norms are untouched: the acoustic representation is inherited from the base.

Audio should be 16 kHz mono.