Mixtral-tiny-GPTQ / config.json
TheBloke's picture
GPTQ model commit
9f35ed9
{
"_name_or_path": "/workspace/process/hf-internal-testing_mixtral-tiny/source",
"architectures": [
"MixtralForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": 2,
"hidden_act": "silu",
"hidden_size": 1024,
"initializer_range": 0.02,
"intermediate_size": 3584,
"max_position_embeddings": 131072,
"model_type": "mixtral",
"num_attention_heads": 32,
"num_experts_per_tok": 2,
"num_hidden_layers": 2,
"num_key_value_heads": 8,
"num_local_experts": 8,
"output_router_logits": false,
"pad_token_id": 0,
"pretraining_tp": 1,
"rms_norm_eps": 1e-05,
"rope_theta": 10000.0,
"router_aux_loss_coef": 0.001,
"sliding_window": 4096,
"tie_word_embeddings": false,
"torch_dtype": "float16",
"transformers_version": "4.36.0.dev0",
"use_cache": true,
"vocab_size": 32000,
"quantization_config": {
"bits": 4,
"group_size": 128,
"damp_percent": 0.1,
"desc_act": true,
"sym": true,
"true_sequential": true,
"model_name_or_path": null,
"model_file_base_name": "model",
"quant_method": "gptq"
}
}