Cgrandjean

Error Kernel crash with GRPO

5

Hello guys, Thanks for the framework which is really cool for all the implemented methods. Now , I wanted to try GRPO . I took an A100-80G GPU and used...

GRPO merging Adapters+ re-training not working

6

Hello, I got those version for libraries: ``` bitsandbytes==0.45.3 deepspeed==0.16.4 transformers==4.49.0 unsloth @ git+https://github.com/unslothai/unsloth.git@5d0ee525c1b6a3522f64ad9722249ae34b584555 unsloth_zoo @ git+https://github.com/unslothai/unsloth-zoo.git@f1f367b871f687dfd8d1c2a1c483f3b6b8142ccb vllm==0.7.2 ``` I recently trained a model with GRPO using Lora adapter: ```...

invalid /' allowed in the parser

4

Hello, I'm using the library and thanks for it. Im using the last version(0.10.9 to this date). I recently encountered a problem : ``` from lmformatenforcer import JsonSchemaParser,CharacterLevelParserConfig from lmformatenforcer.integrations.transformers...

Why are \n generated in the output

1

Hello guys, I would like to train the model with format enforced output. I plan to mask the forced tokens to avoid learning this part. But i see a lot...

[Question]: having multiple workspace at the same time dynamically generated

7

### Do you need to ask a question? - [x] I have searched the existing question and discussions and this question is not already answered. - [x] I believe this...