sgl-project
/

DeepSeek-V4-Flash-FP8

Model card Files Files and versions

ispobock commited on 13 days ago

Commit

8f4e2e3

·

verified ·

1 Parent(s): 9befe99

Update README.md

Files changed (1) hide show

README.md +1 -2

README.md CHANGED Viewed

@@ -13,8 +13,7 @@ FP8 re-packaging of [`deepseek-ai/DeepSeek-V4-Flash`](https://huggingface.co/dee
 Model architecture, tokenizer, chat template, and reference `encoding/` are
 unchanged from the base repo. No fine-tuning, no retraining — weights only.
-Shard count unchanged (46). Size: ~149 GB → ~274 GB.
 ## License
-MIT — see `LICENSE`. Copyright © 2023 DeepSeek.

 Model architecture, tokenizer, chat template, and reference `encoding/` are
 unchanged from the base repo. No fine-tuning, no retraining — weights only.
 ## License
+MIT — see `LICENSE`. Copyright © DeepSeek.