| { |
| "dataset": "s33c67-10m.parquet", |
| "tokenizer": "s33c67-10m-bpe", |
| "run_name": "recgpt-10m-submission", |
| "seed": 0, |
| "data_seed": 0, |
| "microbatch_tok": 32768, |
| "total_batch_tok": 32768, |
| "sequence_len": 256, |
| "epochs": 10, |
| "checkpoint_track": "strict-small", |
| "max_tokens": -1, |
| "lr_embed": 0.005, |
| "lr_block": 0.02, |
| "min_lr": 0.0, |
| "wd_adam": 0.005, |
| "wd_muon": 0.1, |
| "adam_beta1": 0.9, |
| "adam_beta2": 0.997, |
| "muon_momentum": 0.95, |
| "warmup_ratio": 0.0, |
| "cooldown_ratio": 0.2, |
| "max_grad_norm": 2.0, |
| "nl_mult": 0.01, |
| "nl_depth": 2, |
| "nl_hidden": -1, |
| "nl_intermediate": 5120, |
| "nl_lr": 0.004, |
| "nl_wd": 0.01, |
| "nl_momentum": 0.95, |
| "torch_compile": true, |
| "use_wandb": true, |
| "wandb_project": "bblm26-recgpt", |
| "log_every": 10, |
| "model_config": { |
| "transformers_version": "5.9.0", |
| "architectures": [ |
| "RecGPTForCausalLM" |
| ], |
| "output_hidden_states": false, |
| "return_dict": true, |
| "dtype": "float32", |
| "chunk_size_feed_forward": 0, |
| "is_encoder_decoder": false, |
| "id2label": { |
| "0": "LABEL_0", |
| "1": "LABEL_1" |
| }, |
| "label2id": { |
| "LABEL_0": 0, |
| "LABEL_1": 1 |
| }, |
| "problem_type": null, |
| "_name_or_path": "", |
| "pad_token_id": 0, |
| "tie_word_embeddings": false, |
| "vocab_size": 32768, |
| "hidden_size": 768, |
| "embedding_size": 192, |
| "head_dim": 64, |
| "num_heads": 12, |
| "intermediate_size": 12288, |
| "recursive_depth": 16, |
| "max_position_embeddings": 1024, |
| "is_decoder": true, |
| "use_cache": false, |
| "auto_map": { |
| "AutoConfig": "modeling_recgpt.RecGPTConfig", |
| "AutoModelForCausalLM": "modeling_recgpt.RecGPTForCausalLM" |
| }, |
| "model_type": "recgpt", |
| "output_attentions": false |
| } |
| } |
