| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
Add Benchmarking and Fine-Tuning Support for ZenFlow (#982) * Add benchmark scripts and README for ZenFlow - Introduced `zf_benchmark.py` for model offloading benchmarking with DeepSpeed. - Added `output_table.py` to parse and display benchmark results in a tabular format. - Created `run_benchmark.sh` to automate benchmark runs with various configurations. Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> * Add Llama-2 fine-tuning scripts and configuration for ZenFlow - Introduced `finetune_llama.py` for fine-tuning the Llama-2 model using DeepSpeed and ZenFlow. - Added `finetune_llama.sh` for automated training setup with environment variables and DeepSpeed command. - Added `zf_config.json` example for DeepSpeed configuration with ZenFlow optimizations. Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> Co-authored-by: Yusen Wu <xrn4ub@virginia.edu> * Add explanation tips for interpreting benchmark results in README Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> * Add guidance on step/latency interpretation Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> --------- Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> Co-authored-by: Yusen Wu <xrn4ub@virginia.edu> | 1 年前 | |
Fix README for LLaMA-2 fine-tuning with ZenFlow. (#987) Signed-off-by: Tingfeng Lan <erc8gx@virginia.edu> | 1 年前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 1 年前 | ||
| 1 年前 |