-
Restore original mistral3 config for FP8 loading
-
Reshard to 10GB chunks (all tensors preserved)
-
Add cleaned shard model-00001-of-00013.safetensors
-
Remove old shard model-00002-of-00002.safetensors
-
Remove old shard model-00001-of-00002.safetensors
-
Fix: lm_head + dequant FP8->bf16 + scale slicing + config promotion
-
Upload folder using huggingface_hub
-
Upload folder using huggingface_hub
-
Upload folder using huggingface_hub
-
Upload folder using huggingface_hub
-
Upload folder using huggingface_hub
-
initial commit