Ben Shoham Ofir

CPLLM: Clinical Prediction with Large Language Models https://arxiv.org/abs/2309.11295 CPLLM is a novel method that involves fine-tuning a pre-trained Large Language Model for clinical disease prediction, surpasses the state-of-the-art method (Med-BERT)...

ValueError: DeepSpeed Zero-3 is not compatible with `low_cpu_mem_usage=True` or with passing a `device_map`.

7

Hi, I tried to load deepspeed zero3 (large model, I need to split his weights) on multi-gpus. However, when launching zero3 with accelerate I got the following error: ``` ssh://benshoho@/home/benshoho/.conda/envs/accelerate_venv/bin/python...

Assertion `index >= -sizes[i] && index < sizes[i] && "index out of bounds"` failed.

9

### System Info ```Shell torch==2.2.1, tokenizers==0.15.1, transformers==4.38.1, accelerate==0.26.1 linux. 3 gpus of rtx600. CUDA Version: 12.1 accelerate env result: - `Accelerate` version: 0.26.1 - Platform: Linux-3.10.0-1160.90.1.el7.x86_64-x86_64-with-glibc2.17 - Python version: 3.10.9...

Hello, I tried to add a new metric to an existing multiple-choice task, but it seems that the metric was not added. I edited MedQA: ``` task: medqa_4options dataset_path: GBaker/MedQA-USMLE-4-options-hf...

[BFCL] Is there support for running BFCL evaluation with GPT-OSS?

8

Hi, Is there support for running BFCL evaluation with GPT-OSS? For example, https://huggingface.co/openai/gpt-oss-120b

DPO Training Fails with Tool Role Messages - KeyError: 'tool'

1

### Please check that this issue hasn't been reported before. - [x] I searched previous [Bug Reports](https://github.com/axolotl-ai-cloud/axolotl/labels/bug) didn't find any similar reports. ### Expected Behavior # DPO Training Fails with...