File size: 995 Bytes
62dca4c | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 | #!/bin/bash
ROOT_DIR=/workspace/hanrui/junquan/SpecForge
export PYTHONPATH=$ROOT_DIR:$PYTHONPATH
export PATH=/workspace/hanrui/specforge/bin:$PATH
export TORCHINDUCTOR_CACHE_DIR=$ROOT_DIR/cache/compiled_kernels
export SPECFORGE_DATA_NUM_PROC=16
export PYTHONUNBUFFERED=1
export TRANSFORMERS_OFFLINE=1
NUM_GPUS=${1:-8}
/workspace/hanrui/specforge/bin/python3 -m torch.distributed.run \
--standalone \
--nproc_per_node $NUM_GPUS \
$ROOT_DIR/scripts/eval_dflash_lora.py \
--model-path /workspace/models/Qwen3-8B \
--ckpt-dir $ROOT_DIR/outputs/qwen3-8b-dflash-lora/epoch_2_step_218500 \
--data-path /workspace/hanrui/datasets/Nemotron-CodeAlpaca-qwen3-8b-800K \
--lora-config $ROOT_DIR/configs/qwen3-8b-dflash-lora.json \
--block-size 16 \
--max-length 2048 \
--batch-size 1 \
--attention-backend flex_attention \
--lm-head-chunk-size 256 \
--chat-template qwen \
--log-interval 50 \
--cache-dir $ROOT_DIR/cache \
--dist-timeout 120
|