poolside
/

Laguna-M.1-FP8

Text Generation

Mixture of Experts

compressed-tensors

Model card Files Files and versions

Model card: variant-specific rewrite (match XS.2-FP8); add LICENSE.md

#2

by joerowell - opened 14 days ago

base: refs/heads/main

←

from: refs/pr/2

Discussion Files changed

initial commit12423dfd

Add files using upload-large-folder toola7228e8c

Add files using upload-large-folder toole73ae897

Push corrected vLLM image (rope_parameters guard + parser registrations)b01da1d2

Add files using upload-large-folder toolac83b476

Upload image/LOAD.md with huggingface_hub0c2b21ca

Upload image/laguna-vllm-v0.19.0-overlay.tar.gz with huggingface_hub853c0cde

Upload chat_template.jinjaf81f3228

Update overlay image with H100 fp8_w8a8 fused-MoE tuned config (E=256,N=512)cd820de8

Update README.mde5de837c

Bump max_position_embeddings to 256K; double yarn rope factor to 64778ee84d

Add mlp_only_layers, decoder_sparse_step, norm_topk_prob; set sliding_window=0 for vLLM 0.21.0 native Laguna42cd8b8b

update sampling parameters to match evalsedcd8127

Update README.md (#1)361e2d36

Update README.md49c9a7c8

Model card: variant-specific rewrite (match XS.2-FP8); add LICENSE.md5fded70a

Poolside org 14 days ago

No description provided.

Fix vLLM (#41129) and TRT-LLM (#13559, #15110) references04ec8c3d

Add FP8-KV-cache vLLM >= 0.22.0 caveat (vllm#42650)8dcc5c7a

Remove FP8-KV-cache caveat (not applicable to Laguna M.1)a5a0c30a

joerowell changed pull request status to merged 14 days ago

joerowell deleted the refs/pr/2 ref 14 days ago

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