MERaLiON-2-3B-RotorQuant-MLX-2bit / processor_config.json
majentik's picture
Add MLX 2-bit quantized model with KV cache compression
0f832ed verified
raw
history blame contribute delete
281 Bytes
{
"auto_map": {
"AutoProcessor": "processing_meralion2.MERaLiON2Processor"
},
"do_normalize": true,
"fixed_speech_embeds_length": 100,
"processor_class": "MERaLiON2Processor",
"speech_token_index": 255999,
"time_duration_limit": 300,
"whisper_chunk_size": 30
}