| { |
| "datasets": [ |
| { |
| "name": "everyday_conversations", |
| "repo_id": "HuggingFaceTB/everyday-conversations-llama3.1-2k", |
| "format": "messages_all_turns", |
| "train_split": "train_sft", |
| "validation_split": "test_sft", |
| "min_turns": 2, |
| "max_turns": null, |
| "max_message_chars": 6000, |
| "use_base_chat_template": true |
| } |
| ], |
| "model": { |
| "base_model_name": "google/t5gemma-l-l-prefixlm-it", |
| "initial_model_path": "BRlkl/samantha-pretrain256newfixed", |
| "dtype": "bfloat16", |
| "attn_implementation": "sdpa", |
| "magicnorm_eps": 1e-06, |
| "z_slots": 256, |
| "num_time_tokens": 0, |
| "use_explicit_time_features": false, |
| "gate_attention_heads": 4, |
| "max_observation_tokens": 1024, |
| "max_decoder_tokens": 1024, |
| "thought_loop_proposal_mode": "observation_hidden_compression", |
| "preserve_observation_encoder_manifold": true, |
| "observation_encoder_use_state_context": true, |
| "latent_attention_mask_mode": "full", |
| "initial_update_gate_bias": 6.0 |
| }, |
| "training": { |
| "seed": 17, |
| "num_workers": 2, |
| "gradient_checkpointing": true, |
| "mixed_precision": "bf16", |
| "max_grad_norm": 1.0, |
| "weight_decay": 0.01, |
| "backbone_learning_rate": 5e-06, |
| "new_module_learning_rate": 0.0001, |
| "adam_beta1": 0.9, |
| "adam_beta2": 0.95, |
| "adam_epsilon": 1e-08, |
| "fused_adamw": true, |
| "freeze_gate_head": true, |
| "log_every_steps": 1, |
| "eval_every_steps": 25, |
| "checkpoint_every_steps": 250, |
| "max_train_examples": null, |
| "max_validation_examples": null, |
| "freeze_decoder_first_fraction": 1.0, |
| "probe_extra_new_tokens": 16 |
| }, |
| "phase": { |
| "micro_batch_size": 16, |
| "eval_batch_size": 16, |
| "gradient_accumulation_steps": 4, |
| "num_train_epochs": 1, |
| "warmup_ratio": 0.03, |
| "shuffle_train": true |
| }, |
| "cache": { |
| "preprocessed_root": "cache/preprocessed_pre_sft_multiturn" |
| }, |
| "paths": { |
| "run_root": "runs_pre_sft_multiturn", |
| "export_root": "exports_multiturn" |
| }, |
| "inference": { |
| "format": "alpaca_single_turn", |
| "use_base_chat_template": true |
| }, |
| "wandb": { |
| "enabled": true, |
| "project": "samantha-pre-sft", |
| "run_name": "t5gemma2-thoughtloop-pre-sft-everyday-all-turns" |
| }, |
| "hub": { |
| "model_repo_id": "BRlkl/test_multiturn", |
| "private": false |
| } |
| } |
