NullSense/Nanbeige4.2-3B-FP8-Dynamic at main

Nanbeige4.2-3B-FP8-Dynamic

5.21 GB

Ctrl+K

Ctrl+K

1 contributor

History: 21 commits

NullSense's picture

Refresh EAGLE-3 references for the thinking-aware v2 head (2026-07-24 retrain); concurrency section re-measured (501/878/1017 tok/s, ITL 6.2-7.6ms)

4b8767c verified about 6 hours ago

assets
Refresh EAGLE-3 references for the thinking-aware v2 head (2026-07-24 retrain); concurrency section re-measured (501/878/1017 tok/s, ITL 6.2-7.6ms) about 6 hours ago
vllm_plugin
Update bundled plugin: looped-target Eagle3 draft fixes (layer-name offset + target_layer_count stamp) 1 day ago
.gitattributes
1.7 kB
Add concurrency-scaling perf chart (GuideLLM, real data) about 17 hours ago
README.md
14.7 kB
Refresh EAGLE-3 references for the thinking-aware v2 head (2026-07-24 retrain); concurrency section re-measured (501/878/1017 tok/s, ITL 6.2-7.6ms) about 6 hours ago
chat_template.jinja
8.36 kB
Upload folder using huggingface_hub 1 day ago
config.json
3.29 kB
Upload folder using huggingface_hub 1 day ago
configuration_nanbeige.py
20.7 kB
Upload folder using huggingface_hub 1 day ago
generation_config.json
187 Bytes
Upload folder using huggingface_hub 1 day ago
model.safetensors
5.19 GB
xet

Upload folder using huggingface_hub 1 day ago
modeling_nanbeige.py
123 kB
Upload folder using huggingface_hub 1 day ago
recipe.yaml
176 Bytes
Upload folder using huggingface_hub 1 day ago
tokenizer.json
18.5 MB
xet

Upload folder using huggingface_hub 1 day ago
tokenizer_config.json
455 Bytes
Upload folder using huggingface_hub 1 day ago