| { | |
| "activation_function": "swiglu", | |
| "auto_map":{ | |
| "AutoConfig": "configuration_srv1.SRV1Config", | |
| "AutoModelForCausalLM": "modeling_srv1_tp.SRV1ForCausalLMParallel" | |
| }, | |
| "bos_token_id": 13, | |
| "eos_token_id": 13, | |
| "hidden_act": "silu", | |
| "hidden_size": 2304, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 6144, | |
| "max_position_embeddings": 8192, | |
| "num_attention_heads": 18, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 18, | |
| "pad_token_id": 0, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": {"type":"linear", "factor": 2.0 }, | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.31.0", | |
| "use_cache": true, | |
| "torch_dtype": "bfloat16", | |
| "vocab_size": 96000 | |
| } | |