torchchat icon indicating copy to clipboard operation
torchchat copied to clipboard

Run PyTorch LLMs locally on servers, desktop and mobile

Results 143 torchchat issues
Sort by recently updated
recently updated
newest added

https://github.com/pytorch/torchchat/actions/runs/9161687849/job/25187114502?pr=831 January 16, 2019 --------------------------------------- Accept? (y/N): Skipping following packages as the license is not accepted: Google APIs Intel x86_64 Atom System Image The following packages can not be installed...

Discussion about PyTorch/TorchChat performance on ARM (more specifically on Apple Silicon) is tracked discussed here In particular, `torch.compile` generates slower code than eager on the platform, see https://github.com/pytorch/pytorch/issues/125734 as a...

> "UserWarning: Mutation on a buffer in the model is detected. ExecuTorch assumes buffers that are mutated in the graph have a meaningless initial state, only the shape and dtype...

the evaluation.md file contains instructions that are not reproducible with torchchat/scripts/run-docs evaluation At present that's fixed by `exit 0` before the test commences. The reported error is missing a checkpoint,...

I want to increadse the number of tries as a user if I'm on a bad connection and let it download unattended. I've tried to download stuff from other countries,...

Remove `-l 2` and `-l 3` flags and auto-detect model architecture from tokenizer class. Issue warning if user supplied -v does not match the tokenizer.model inferred vocab size.

I get it -- maybe the error is `aten::_weight_int4pack_mm(Tensor self, Tensor mat2, int qGroupSize, Tensor qScaleAndZeros) -> Tensor' to its out variant with error: 'SchemaKind.out variant of operator aten::_weight_int4pack_mm can't...

We should not block improvements to iOS build on a complete refactor to re-use the existing build. That being said this should be a high priority item for future. Also,...

This is running a back-level version of macos but may be supported by PyTorch? maybe treat it as PT improvement opportunity, not Torchchat launch blocker? (x-ref: https://github.com/pytorch/pytorch/issues/125856) https://github.com/pytorch/torchchat/actions/runs/9006021620/job/24742618846?pr=723 ld: warning:...

``` % python3 torchchat.py generate --dtype float16 --checkpoint-path ~/git/meta-llama/llama/llama-2-7b/consolidated.00.pth --tokenizer-path ~/git/meta-llama/llama/tokenizer.model /Users/nshulga/Library/Python/3.9/lib/python/site-packages/urllib3/__init__.py:35: NotOpenSSLWarning: urllib3 v2 only supports OpenSSL 1.1.1+, currently the 'ssl' module is compiled with 'LibreSSL 2.8.3'. See: https://github.com/urllib3/urllib3/issues/3020...