Text Generation
Transformers
Safetensors
English
mixtral
Generated from Trainer
axolotl
conversational
Inference Endpoints
text-generation-inference
8-bit precision
exl2
FuturisticVibes's picture
Upload folder using huggingface_hub
e1fe2bf verified
{
"_name_or_path": "mistralai/Mixtral-8x22B-v0.1",
"architectures": [
"MixtralForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 1,
"eos_token_id": 32000,
"hidden_act": "silu",
"hidden_size": 6144,
"initializer_range": 0.02,
"intermediate_size": 16384,
"max_position_embeddings": 65536,
"model_type": "mixtral",
"num_attention_heads": 48,
"num_experts_per_tok": 2,
"num_hidden_layers": 56,
"num_key_value_heads": 8,
"num_local_experts": 8,
"output_router_logits": false,
"rms_norm_eps": 1e-05,
"rope_theta": 1000000,
"router_aux_loss_coef": 0.001,
"router_jitter_noise": 0.0,
"sliding_window": null,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.40.2",
"use_cache": false,
"vocab_size": 32002,
"quantization_config": {
"quant_method": "exl2",
"version": "0.1.4",
"bits": 8.0,
"head_bits": 8,
"calibration": {
"rows": 100,
"length": 2048,
"dataset": "(default)"
}
}
}