GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
amzilmustapha
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
amzilmustapha
Model Tree
CohereLabs/cohere-transcribe-arabic-07-2026
Input Modalities
Output Modalities
Supported Functionality
The model was fine-tuned using Parameter-Efficient Fine-Tuning (PEFT) with LoRA and 8-bit quantization.
You can use this adapter by loading the base model and merging it with this LoRA repository.
python
import librosaimport torchfrom transformers import AutoFeatureExtractor, AutoTokenizer, CohereAsrForConditionalGenerationfrom peft import PeftModelbase_model_id = "CohereLabs/cohere-transcribe-arabic-07-2026"adapter_repo = "amzilmustapha/cohere-darija-lora-10-7-26"hf_token = "YOUR_HF_TOKEN" # Required for the gated base model# Load processorsfeature_extractor = AutoFeatureExtractor.from_pretrained(base_model_id, trust_remote_code=True, token=hf_token)tokenizer = AutoTokenizer.from_pretrained(base_model_id, trust_remote_code=True, token=hf_token)# Load base modelbase_model = CohereAsrForConditionalGeneration.from_pretrained(base_model_id,device_map="auto",trust_remote_code=True,token=hf_token)# Apply LoRA Adaptermodel = PeftModel.from_pretrained(base_model, adapter_repo, token=hf_token)# Process Audioaudio_file = "your_darija_audio.wav"audio_array, sampling_rate = librosa.load(audio_file, sr=16000)inputs = feature_extractor(audio_array, sampling_rate=16000, return_tensors="pt")input_features = inputs.input_features.to(model.device)# Generate Transcriptionwith torch.no_grad():generated_ids = model.generate(inputs=input_features, max_new_tokens=100)transcription = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]print(transcription)