| { | |
| "_notes": "Extra training-only keys kept below; HF ignores them but they\u2019re here for provenance.", | |
| "architectures": [ | |
| "MyCustomModelForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "blahblahtron_1_1B.HFWrapperConfig", | |
| "AutoModelForCausalLM": "blahblahtron_1_1B.MyCustomModelForCausalLM" | |
| }, | |
| "bos_token_id": 50256, | |
| "dropout": 0.1, | |
| "embedding_dim": 2048, | |
| "eos_token_id": 50256, | |
| "ffn_dim_multiplier": 4.0, | |
| "learning_rate": 0.0003, | |
| "context_length": 1024, | |
| "model_type": "shallowmind", | |
| "num_heads": 16, | |
| "num_kv_heads": 4, | |
| "num_layers": 22, | |
| "pad_token_id": 50256, | |
| "rms_norm_eps": 1e-06, | |
| "tokenizer_name_or_path": "gpt2", | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.55.2", | |
| "use_flash_attention_2": true, | |
| "vocab_size": 50257, | |
| "weight_decay": 0.1 | |
| } | |