lightseekorg
/

kimi-k2.5-eagle3-mla

Text Generation

speculative-decoding

Model card Files Files and versions

Convert weights to bf16; add max_position_embeddings

#2

by yueweng - opened Apr 30

base: refs/heads/main

←

from: refs/pr/2

Discussion Files changed

Apr 30

•

Convert all 17 tensors from fp16 to bf16 using round-to-nearest-ties-to-even rounding.
Add max_position_embeddings=262144 to config.json.

Convert weights to bf16; add max_position_embeddings11318080

Apr 30

Superseded by a re-submitted PR with corrected Signed-off-by email (yue@lightseek.org).

yueweng changed pull request status to closed Apr 30

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment