diff options
| author | YurenHao0426 <blackhao0426@gmail.com> | 2026-01-27 09:57:37 -0600 |
|---|---|---|
| committer | YurenHao0426 <blackhao0426@gmail.com> | 2026-01-27 09:57:37 -0600 |
| commit | dc801c07cf38b0c495686463e6ca6f871a64440e (patch) | |
| tree | 599f03114775921dbc472403c701f4a3a8ea188a /collaborativeagents/training/train_sft_resume.sbatch | |
| parent | e43b3f8aa36c198b95c1e46bea2eaf3893b13dc3 (diff) | |
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 <noreply@anthropic.com>
Diffstat (limited to 'collaborativeagents/training/train_sft_resume.sbatch')
| -rw-r--r-- | collaborativeagents/training/train_sft_resume.sbatch | 44 |
1 files changed, 44 insertions, 0 deletions
diff --git a/collaborativeagents/training/train_sft_resume.sbatch b/collaborativeagents/training/train_sft_resume.sbatch new file mode 100644 index 0000000..e6a868b --- /dev/null +++ b/collaborativeagents/training/train_sft_resume.sbatch @@ -0,0 +1,44 @@ +#!/bin/bash +#SBATCH --job-name=sft_resume +#SBATCH --account=bfqt-delta-gpu +#SBATCH --partition=gpuH200x8 +#SBATCH --nodes=1 +#SBATCH --ntasks=1 +#SBATCH --cpus-per-task=32 +#SBATCH --gres=gpu:4 +#SBATCH --mem=256G +#SBATCH --time=1:00:00 +#SBATCH --output=/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/sft_resume_%j.out +#SBATCH --error=/projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/sft_resume_%j.err + +echo "=== SFT Training Resume from checkpoint-100 (H200) ===" +date +nvidia-smi --query-gpu=index,name,memory.total --format=csv + +cd /projects/bfqt/users/yurenh2/ml-projects/personalization-user-model/collaborativeagents/training +source /u/yurenh2/miniforge3/etc/profile.d/conda.sh +conda activate eval +export HF_HOME=/projects/bfqt/users/yurenh2/hf_cache/huggingface +export WANDB_MODE=offline + +echo "Model: llama-3.1-8b-instruct" +echo "Resuming from: /work/nvme/bfqt/yurenh2/sft_checkpoints/checkpoint-100" +echo "Output: /work/nvme/bfqt/yurenh2/sft_checkpoints" + +# Verify checkpoint exists +if [ -d "/work/nvme/bfqt/yurenh2/sft_checkpoints/checkpoint-100" ]; then + echo "Checkpoint verified!" +else + echo "ERROR: Checkpoint not found!" + exit 1 +fi + +echo "" +echo "Starting LLaMA-Factory SFT training resume with DeepSpeed ZeRO-3..." + +# Run with LLaMA-Factory CLI + DeepSpeed ZeRO-3 +FORCE_TORCHRUN=1 llamafactory-cli train llama_factory_resume_config.yaml + +echo "" +echo "SFT Training complete!" +date |
