Or4cl3-1's picture
Update config.json
9523b00 verified
raw
history blame contribute delete
No virus
778 Bytes
{
"_name_or_path": "/workspace/models/Or4cl3-1/Cognitive-Agent-Gemma_7b",
"architectures": [
"CognitiveAgentForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 0,
"eos_token_id": 2,
"hidden_act": "gelu",
"hidden_size": 1024,
"initializer_range": 0.02,
"intermediate_size": 4096,
"max_position_embeddings": 1024,
"model_type": "text-gemma-001",
"num_attention_heads": 16,
"num_hidden_layers": 12,
"num_key_value_heads": 1,
"num_local_experts": 1,
"output_router_logits": false,
"rms_norm_eps": 1e-05,
"rope_theta": 1000000.0,
"router_aux_loss_coef": 0.02,
"sliding_window": null,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.36.0.dev0",
"use_cache": false,
"vocab_size": 50265
}