Automatic Speech Recognition
Transformers
Safetensors
Arabic
English
cohere_asr
arabic
code-switching
decoder-only-finetune
Instructions to use nsa01n/cohere-cs-decoder-full with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use nsa01n/cohere-cs-decoder-full with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="nsa01n/cohere-cs-decoder-full")# pip install -U transformers accelerate # Load model directly from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq processor = AutoProcessor.from_pretrained("nsa01n/cohere-cs-decoder-full") model = AutoModelForSpeechSeq2Seq.from_pretrained("nsa01n/cohere-cs-decoder-full", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download train_meta.json from nsa01n/cohere-cs-decoder-full: direct link, hf CLI and curl.
- Browser
- Download file 1.38 kB
-
https://huggingface.co/nsa01n/cohere-cs-decoder-full/resolve/main/train_meta.json
- Command line
-
hf download hf://nsa01n/cohere-cs-decoder-full/train_meta.json
-
curl -L -o train_meta.json https://huggingface.co/nsa01n/cohere-cs-decoder-full/resolve/main/train_meta.json
1.38 kB
| { | |
| "base_model": "CohereLabs/cohere-transcribe-arabic-07-2026", | |
| "variant": "full", | |
| "method": "decoder_only_finetune", | |
| "target_column": "english", | |
| "dataset": "Ahmed1/cohere-asr-cs", | |
| "effective_batch_size": 32, | |
| "seed": 42, | |
| "bf16": true, | |
| "train_examples": 1900, | |
| "val_examples": 454, | |
| "metrics": { | |
| "train_runtime": 419.5181, | |
| "train_samples_per_second": 27.174, | |
| "train_steps_per_second": 0.858, | |
| "total_flos": 5.372639220298678e+19, | |
| "train_loss": 0.10716057336992688, | |
| "epoch": 6.0 | |
| }, | |
| "freeze": { | |
| "variant": "full", | |
| "decoder_module": "model.decoder", | |
| "decoder_layers_total": 8, | |
| "decoder_layers_trained": [ | |
| 0, | |
| 1, | |
| 2, | |
| 3, | |
| 4, | |
| 5, | |
| 6, | |
| 7 | |
| ], | |
| "train_decoder_io": true, | |
| "output_head": "proj_out", | |
| "io_trainable_params": 35935232, | |
| "layer_trainable_params": 134373376, | |
| "encoder_module": "model.encoder", | |
| "encoder_trainable_params": 0, | |
| "total_params": 2065647872, | |
| "trainable_params": 170308608, | |
| "trainable_pct": 8.245 | |
| }, | |
| "hyperparameters": { | |
| "learning_rate": 8e-05, | |
| "num_train_epochs": 6, | |
| "lr_scheduler_type": "cosine", | |
| "warmup_ratio": 0.03, | |
| "per_device_train_batch_size": 16, | |
| "gradient_accumulation_steps": 2, | |
| "optim": "adamw_bnb_8bit", | |
| "weight_decay": 0.01, | |
| "max_grad_norm": 1.0, | |
| "save_total_limit": 1 | |
| } | |
| } | |