instruct+safety mix SFT (field=messages_cite, 30% safety = 90000 safety + 210000 instruct of 300000; instruct=jkminder/model-raising-pbsft-instruct-300k safety=jkminder/model-raising-pbsft-safety-180k, template=epe-template-nosys, tokenizer=model-dir) on epe-1p-3b-llama3arch-smollm2tok-500B-40n-2048sl-960gbsz-no_bce
4a1e140 verified | { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 1, | |
| "dtype": "bfloat16", | |
| "eos_token_id": 2, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 3072, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 8192, | |
| "max_position_embeddings": 2048, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 24, | |
| "num_hidden_layers": 28, | |
| "num_key_value_heads": 8, | |
| "pad_token_id": 2, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "4.57.6", | |
| "use_cache": true, | |
| "vocab_size": 49280 | |
| } | |