From dc801c07cf38b0c495686463e6ca6f871a64440e Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Tue, 27 Jan 2026 09:57:37 -0600 Subject: Add collaborativeagents module and update gitignore - Add collaborativeagents subproject with adapters, agents, and evaluation modules - Update .gitignore to exclude large binary files (.whl, .tar), wandb logs, and results Co-Authored-By: Claude Opus 4.5 --- .../training/llama_factory_resume_config.yaml | 38 ++++++++++++++++++++++ 1 file changed, 38 insertions(+) create mode 100644 collaborativeagents/training/llama_factory_resume_config.yaml (limited to 'collaborativeagents/training/llama_factory_resume_config.yaml') diff --git a/collaborativeagents/training/llama_factory_resume_config.yaml b/collaborativeagents/training/llama_factory_resume_config.yaml new file mode 100644 index 0000000..627d5a4 --- /dev/null +++ b/collaborativeagents/training/llama_factory_resume_config.yaml @@ -0,0 +1,38 @@ +### LLaMA-Factory SFT Training Config - Resume from checkpoint ### +### For session-level reflection training ### + +### Model +model_name_or_path: /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/models/llama-3.1-8b-instruct + +### Method +stage: sft +do_train: true +finetuning_type: full +deepspeed: ds_z3_config.json + +### Dataset +dataset: sft_reflection +dataset_dir: /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/collaborativeagents/training +template: llama3 +cutoff_len: 4096 + +### Output - Using NVMe storage for checkpoints +output_dir: /work/nvme/bfqt/yurenh2/sft_checkpoints + +### Resume from checkpoint +resume_from_checkpoint: /work/nvme/bfqt/yurenh2/sft_checkpoints/checkpoint-100 + +### Training +per_device_train_batch_size: 1 +gradient_accumulation_steps: 16 +learning_rate: 1.0e-6 +num_train_epochs: 4.0 +lr_scheduler_type: cosine +warmup_ratio: 0.1 +bf16: true +ddp_timeout: 3600 + +### Logging +logging_steps: 10 +save_steps: 300 +save_total_limit: 2 -- cgit v1.2.3