simaai 's Collections

LLMs for Accuracy Benchmarking

Precompiled LLMs with --return_logits for Modalix accuracy benchmarking, with SiMa.ai-published results and HF GPU reference baselines.