Sandeep Das

sandeep85's activity

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

New activity in LocalLLaMA/typed-decisions 1 day ago

K-Decision-9B (ONYULTEA) — 0.818 accuracy (fitted)

2

#67 opened 3 days ago by

ONYULTEA

Add K-Decision-9B (#67) to the fitted table

1

#69 opened 1 day ago by

sandeep85

New activity in LocalLLaMA/typed-decisions 3 days ago

[Leaderboard Submission] nanoGentzen-v3 (4.93M) - 76.75% Accuracy (General) - New Version

3

#59 opened 6 days ago by

Sagicc

Move nanoGentzen to v3 (#59) in the fitted table

1

#66 opened 3 days ago by

sandeep85

New activity in LocalLLaMA/typed-decisions 5 days ago

Submission: ten Rev-gen1 models (general), 150M–25.6B

2

#64 opened 5 days ago by

delip

Add ten Rev-gen1 models (#64) to the fitted table

1

#65 opened 5 days ago by

sandeep85

Interactive explorer for the leaderboard: ranking + Pareto charts, reads this card live

❤️🔥 1

2

#62 opened 5 days ago by

tardellirs

Submission: Qwen/Qwen3.8-27B base model (general, zero-shot): 0.661 acc / 0.590 KL / 0.214 Brier, plus FP8, NVFP4 and Swift 1.5

🔥 1

1

#60 opened 5 days ago by

adcostanza

Submission: skamalj/signpost-d-4b (4.5B, general, zero-shot): 0.712 acc / 0.1615 KL / 0.0853 Brier / 0.0914 ECE

1

#61 opened 5 days ago by

skamalj

Submission: TypeScript/FastBrain-9B (9.4B): 0.7625 acc / 0.1508 KL / 0.0795 Brier (zero-shot) & 0.8070 acc / 0.0816 KL (fitted)

3

#57 opened 6 days ago by

TypeScript

Add signpost-d-4b (#61), Qwen3.8-27B base (#60) and FastBrain-9B fitted (#57)

1

#63 opened 5 days ago by

sandeep85

New activity in LocalLLaMA/typed-decisions 6 days ago

Add FastBrain-9B (#57) to the zero-shot table

1

#58 opened 6 days ago by

sandeep85

Submission: gnumanth/nullspan v1.3.0 (395M 8K tree-isolated encoder, specialist): 0.7900 acc / 0.0760 KL / 0.0420 Brier / 0.1665 ECE

2

#55 opened 7 days ago by

gnumanth

New activity in LocalLLaMA/typed-decisions 7 days ago

Leaderboard submission: roudone/Decider-4B (fine-tuned on train)

6

#36 opened 22 days ago by

roudone

Update Decider-4B (#36) and add nullspan v1.3.0 (#55) to the fitted table

1

#56 opened 7 days ago by

sandeep85

Add d1-2B and d1-2B-OptiQ-4bit (zero-shot)

1

#54 opened 7 days ago by

sandeep85

New activity in mlx-community/Qwen3.5-9B-OptiQ-4bit 7 days ago

Using the MTP head with oMLX Lightning MTP: 3 norms need a +1 shift

1

#2 opened 8 days ago by

d-treegardener

New activity in LocalLLaMA/typed-decisions 8 days ago

Add Sagicc/nanoGentzen-Jev-v3 (4.93M params) — 0.731 acc, 0.150 KL, 0.081 Brier, 0.087 ECE

4

#52 opened 9 days ago by

Sagicc

New activity in LocalLLaMA/typed-decisions 9 days ago

Add nanoGentzen-Jev-v3 (#52) to the zero-shot table

#53 opened 9 days ago by

sandeep85

New activity in CohereLabs/10-2026-qwen-investigation-traces 10 days ago

This repro script only tests forced URL autocomplete, not a reproducible call-home

🔥👍 22

5

#1 opened 11 days ago by

hakanavgin