Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
jjbuck created a comment on an issue on huggingface/transformers
@SunMarc It looks like it was commit `8b3b9b48fcd6bc06bd9c576f1b09266d577db257` [https://github.com/huggingface/transformers/commit/8b3b9b48fcd6bc06bd9c576f1b09266d577db257]. Among other things,...
jjbuck created a comment on an issue on huggingface/transformers
I don't think this is strictly a SageMaker issue. In the time since I first posted this issue, I've reproduced it on an EC2 instance. See below for relevant details. ### Environment details `...
jjbuck opened an issue on huggingface/transformers
Training config that worked with v4.4.6.3 results in OOM error with v4.47.0 when using SFTTrainer
### System Info ``` - `transformers` version: 4.47.0 - Platform: Linux-6.8.0-1015-aws-x86_64-with-glibc2.35 - Python version: 3.12.6 - Huggingface_hub version: 0.26.2 - Safetensors version: 0...
jjbuck opened an issue on huggingface/text-generation-inference
"sharded is not supported for AutoModel" Error When Deploying SageMaker Endpoint For Qwen 2.5 7B Trained via SageMaker
### System Info 1. Docker image: `763104351884.dkr.ecr.us-west-2.amazonaws.com/huggingface-pytorch-tgi-inference:2.1.1-tgi1.4.0-gpu-py310-cu121-ubuntu20.04` 2. Deployment: SageMaker endpoint, usi...
jjbuck opened an issue on huggingface/transformers
Recomputed tensor size does not match when using activation checkpointing in FSDP strategy
### System Info ``` - `transformers` version: 4.46.3 - Platform: Linux-6.8.0-1015-aws-x86_64-with-glibc2.35 - Python version: 3.12.6 - Huggingface_hub version: 0.26.2 - Safetensors version: 0...
jjbuck created a comment on an issue on huggingface/transformers
I recently ran into what appears to be the same issue. If I run with `gradient_checkpointing: true`, I get a warning during training that reads: `activation_checkpointing: "true" # When using FS...
jjbuck created a comment on an issue on microsoft/Phi-3CookBook
For visibility, a contributor to a forked version of the Phi 3 Vision cookbook [suggested](https://github.com/kazuar/Phi3-Vision-ft/issues/2) the following solution, stating "You need to remove the...
jjbuck opened an issue on huggingface/transformers
Changes required to `save_model` for certain models (e.g., Phi 3.5 Vision)
### Feature request This request proposes one of two changes (see **Motivation** for background). 1. Accept a `state_dict` argument in the `Trainer` class's `save_model()` method (https://githu...
jjbuck created a comment on an issue on kazuar/Phi3-Vision-ft
Thank you @2U1! That seems to have worked!
jjbuck opened an issue on microsoft/Phi-3CookBook
RuntimeError When Saving Phi 3.5 Vision Due to Shared Tensors
I’m trying to fine-tune Phi 3.5 Vision using transformers. However, I’m running into an issue trying to save the model during or after training. See below for a minimal reproducible example. This s...
jjbuck created a comment on an issue on kazuar/Phi3-Vision-ft
Ran into the same issue with Phi 3.5 Vision. I posted about this over on the HF forums [https://discuss.huggingface.co/t/runtimeerror-when-saving-phi-3-5-vision-due-to-shared-tensors/116457]. E...
jjbuck created a comment on an issue on huggingface/transformers
I ran into this same thing with https://huggingface.co/microsoft/Phi-3.5-vision-instruct. Hope it's OK to resurrect this thread from September.... Setting `trust_remote_code=True` seems to be ...