Bengaluru-based Polymage Labs and US firm Tenstorrent are set to develop an AI compiler for Tenstorrent's hardware. This collaboration aims to enhance software support for advanced AI chips. Polymage ...
This project was originally developed using Python 3.9, PyTorch 2.1.1, and CUDA 12.1 on Linux. To reproduce the environment, follow these steps: EMAT can be trained for Few-Shot Classification and ...
2025-12-9: Added the LVLLM_MOE_USE_WEIGHT environment variable to support MOE modules using two modes to infer fp8 models LVLLM_MOE_USE_WEIGHT="KEEP": lk_moe inference uses the original weight format ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback