| {"name": "checkpoints/m1_40M_lr1e-3_steps200k_bs8_seqlen2048_python", "dump_dir": "checkpoints/m1_40M_lr1e-3_steps200k_bs8_seqlen2048_python", "seed": 42, "grad_acc_steps": 1, "gc_collect_freq": 1000, "probe_freq": null, "steps": 200000, "data": {"root_dir": "data", "sources": {"m1_python": 1.0}, "batch_size": 8, "seq_len": 2048, "n_views": 2, "seed": 42, "add_bos": true, "add_eos": true, "load_async": true, "prefetch_size": 64, "tokenizer": {"name": "bytes", "path": null}}, "optim": {"lr": 0.001, "weight_decay": 0.1, "epsilon": 1e-15, "beta1": 0.9, "beta2": 0.95, "clip": 1.0, "scheduler": "cosine", "warmup": 1000, "lr_min_ratio": 0.001, "cycle_length": 1.0, "cosine_theta": 1.0, "annealing_step": 1000, "decay_fraction": 0.1, "exp_factor": 0.5}, "model": {"dim": 512, "n_layers": 12, "head_dim": null, "n_heads": 8, "n_kv_heads": null, "ffn_dim_multiplier": null, "multiple_of": 256, "norm_eps": 1e-05, "rope_theta": 10000.0, "init_base_std": null, "init_std_factor": "disabled", "max_seqlen": 2048, "seed": 42, "vocab_size": 256, "weight_tying": false, "sliding_window": null}, "distributed": {"dp_shard": 1, "dp_replicate": 8, "tp_size": 1, "selective_activation_checkpointing": false, "compile": true, "fsdp_type": "full_shard", "model_dtype": "bf16", "float8_recipe": null, "float8_filter": "layers\\.[0-9]+\\.", "matmul_allow_tf32": false, "detect_anomaly": false, "compile_cache_size_limit": 8, "spawn_method": "forkserver"}, "env": {"MKL_SERVICE_FORCE_INTEL": "GNU", "OMP_NUM_THREADS": "1", "MKL_NUM_THREADS": "1", "ENABLE_INTRA_NODE_COMM": "1", "TORCH_NCCL_AVOID_RECORD_STREAMS": "1", "NCCL_IB_TIMEOUT": "22", "NCCL_DEBUG": "INFO", "TORCH_NCCL_ASYNC_ERROR_HANDLING": "1"}, "checkpoint": {"dump": {"every": 10000, "keep": 5}, "eval": {"every": 1000, "keep": 0}, "path": "checkpoints/m1_40M_lr1e-3_steps200k_bs8_seqlen2048_python/checkpoints", "init_ckpt_path": null, "continue_training_from_init": false}, "profiling": {"run": false, "trace_folder": "profiling", "mem_warmup": 100, "mem_steps": 2, "profile_warmup": 102, "profile_steps": 2}, "logging": {"freq": 10, "acc_freq": null, "wandb": {"job_type": null, "dir": null, "project": "m1-byte-model-training", "entity": "linzhengisme-hku", "name": "checkpoints/m1_40M_lr1e-3_steps200k_bs8_seqlen2048_python"}}, "async_eval_gpus": null, "eval": null} |