| { | |
| "architectures": [ | |
| "Olmo2ForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "eos_token_id": 100257, | |
| "hidden_act": "silu", | |
| "hidden_size": 5120, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 13824, | |
| "max_position_embeddings": 32768, | |
| "model_type": "olmo2", | |
| "num_attention_heads": 40, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 40, | |
| "pad_token_id": 100277, | |
| "quantization_config": { | |
| "autoround_version": "0.6.0", | |
| "batch_size": 2, | |
| "bits": 4, | |
| "data_type": "int", | |
| "gradient_accumulate_steps": 4, | |
| "group_size": 128, | |
| "packing_format": "auto_round", | |
| "quant_method": "auto-round", | |
| "seqlen": 512, | |
| "sym": false | |
| }, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 4134231.132028111, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.55.0", | |
| "use_cache": true, | |
| "vocab_size": 100352 | |
| } | |