glaiveai/glaive-function-calling-v2-small

1{2  "_name_or_path": "replit/replit-code-v1-3b",3  "architectures": [4    "MPTForCausalLM"5  ],6  "attn_config": {7    "alibi": true,8    "alibi_bias_max": 8,9    "attn_impl": "torch",10    "attn_pdrop": 0,11    "attn_type": "multihead_attention",12    "attn_uses_sequence_id": false,13    "clip_qkv": null,14    "prefix_lm": false,15    "qk_ln": false,16    "softmax_scale": null17  },18  "auto_map": {19    "AutoConfig": "configuration_mpt.MPTConfig",20    "AutoModelForCausalLM": "modeling_mpt.MPTForCausalLM"21  },22  "d_model": 2560,23  "emb_pdrop": 0,24  "embedding_fraction": 1.0,25  "eos_token_id": 1,26  "expansion_ratio": 4,27  "init_config": {28    "emb_init_std": null,29    "emb_init_uniform_lim": null,30    "fan_mode": "fan_in",31    "init_div_is_residual": true,32    "init_gain": 0,33    "init_nonlinearity": "relu",34    "init_std": 0.02,35    "name": "kaiming_normal_",36    "verbose": 037  },38  "init_device": "cpu",39  "learned_pos_emb": true,40  "logit_scale": null,41  "max_seq_len": 2048,42  "model_type": "mpt",43  "n_heads": 32,44  "n_layers": 32,45  "no_bias": true,46  "norm_type": "low_precision_layernorm",47  "resid_pdrop": 0,48  "tokenizer_name": "replit/replit-code-v1-3b",49  "torch_dtype": "float32",50  "transformers_version": "4.28.0",51  "use_cache": true,52  "verbose": 0,53  "vocab_size": 3276854}55