| { |
| "_name_or_path": "/home/jovyan/model", |
| "activation_function": "silu", |
| "architectures": [ |
| "ExpandedJetMoEForCausalLM" |
| ], |
| "bias": true, |
| "bos_token_id": 1, |
| "eos_token_id": 2, |
| "ffn_hidden_size": 5632, |
| "glu": true, |
| "hidden_size": 2048, |
| "initializer_range": 0.01, |
| "kv_channels": 128, |
| "layer_norm_epsilon": 1e-05, |
| "max_position_embeddings": 4096, |
| "model_type": "expandedjetmoe", |
| "moe_attn_num_experts": 8, |
| "moe_mlp_num_experts": 9, |
| "moe_top_k": 2, |
| "n_positions": 4096, |
| "num_attention_heads": 32, |
| "num_hidden_layers": 24, |
| "num_key_value_heads": 16, |
| "num_layers": 24, |
| "rms_norm_eps": 1e-05, |
| "rope_theta": 10000.0, |
| "rotary_percent": 1.0, |
| "router_loss_coef": 1e-05, |
| "router_loss_type": "reg_loss", |
| "torch_dtype": "bfloat16", |
| "transformers_version": "4.40.2", |
| "use_cache": true, |
| "vocab_size": 32000 |
| } |
|
|