Osaurus-AI

Osaurus-AI's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

New activity in JANGQ-AI/GLM-5.3-FP8 23 days ago

fix: default clear_thinking=true in chat template + add repetition_penalty=1.1 default — mitigates known GLM-5.3 reasoning-loop degeneration on multi-turn agent workloads (SGLang #36669, opencode #45533). Upstream Z.ai default `clear_thinking=false` replays prior reasoning into subsequent turns, priming a decode attractor. Z.ai's own docs recommend `clear_thinking=true` for chat scenarios — this bakes it in.

#1 opened 23 days ago by

Osaurus-AI

New activity in JANGQ-AI/GLM-5.3-Flash-W4A16 23 days ago

fix: default clear_thinking=true in chat template + add repetition_penalty=1.1 default — mitigates known GLM-5.3 reasoning-loop degeneration on multi-turn agent workloads (SGLang #36669, opencode #45533). Upstream Z.ai default `clear_thinking=false` replays prior reasoning into subsequent turns, priming a decode attractor. Z.ai's own docs recommend `clear_thinking=true` for chat scenarios — this bakes it in.

#1 opened 23 days ago by

Osaurus-AI

New activity in JANGQ-AI/GLM-5.3-W4A16 23 days ago

fix: default clear_thinking=true in chat template + add repetition_penalty=1.1 default — mitigates known GLM-5.3 reasoning-loop degeneration on multi-turn agent workloads (SGLang #36669, opencode #45533). Upstream Z.ai default `clear_thinking=false` replays prior reasoning into subsequent turns, priming a decode attractor. Z.ai's own docs recommend `clear_thinking=true` for chat scenarios — this bakes it in.

#1 opened 23 days ago by

Osaurus-AI

New activity in JANGQ-AI/GLM-5.3-Flash-JANG-MTP 23 days ago

fix: default clear_thinking=true in chat template + add repetition_penalty=1.1 default — mitigates known GLM-5.3 reasoning-loop degeneration on multi-turn agent workloads (SGLang #36669, opencode #45533). Upstream Z.ai default `clear_thinking=false` replays prior reasoning into subsequent turns, priming a decode attractor. Z.ai's own docs recommend `clear_thinking=true` for chat scenarios — this bakes it in.

#1 opened 23 days ago by

Osaurus-AI

New activity in JANGQ-AI/GLM-5.3-Flash-JANG 23 days ago

fix: default clear_thinking=true in chat template + add repetition_penalty=1.1 default — mitigates known GLM-5.3 reasoning-loop degeneration on multi-turn agent workloads (SGLang #36669, opencode #45533). Upstream Z.ai default `clear_thinking=false` replays prior reasoning into subsequent turns, priming a decode attractor. Z.ai's own docs recommend `clear_thinking=true` for chat scenarios — this bakes it in.

#1 opened 23 days ago by

Osaurus-AI

New activity in dealignai/GLM-5.3-Flash-UNCENSORED-FP8 27 days ago

GGUF?

5

#1 opened 29 days ago by

kabachuha

New activity in OsaurusAI/MiniMax-M2.7-Small-JANGTQ 5 months ago

config.json: per-module overrides + rope_parameters (metadata bug fix)

1

#1 opened 5 months ago by

jangq

New activity in dealignai/Gemma-4-31B-JANG_4M-CRACK 6 months ago

Upload folder using huggingface_hub

#14 opened 6 months ago by

Osaurus-AI

Upload README.md with huggingface_hub

#13 opened 6 months ago by

Osaurus-AI