dealign.ai PRO

dealignai's activity

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

New activity in dealignai/GLM-5.3-UNCENSORED-FP8 3 days ago

GLM-5.3-UNCENSORED vs GLM-5.3-ABLITERATED

🔥 2

13

#6 opened 4 days ago by

levelsio1

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 4 days ago

Community benchmark: 8x DGX Spark GB10 (TP=8/EP=8) - stable 24/24, ~25% prefill slower than official -1017 on the same hardware

6

#7 opened 7 days ago by

jlacroix82

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 6 days ago

HuggingFace is taking down your models. Upload them to pirateface.co

4

#8 opened 6 days ago by

StellarBlue

Sir, could you create a version that runs locally on a MacBook—for instance, a version compatible with LM Studio?

3

#6 opened 7 days ago by

wen-88-QQ

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 11 days ago

this is not needed

👍 1

15

#3 opened 13 days ago by

brainrottrbgtrbtrh

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 12 days ago

VQ Q4K recipe

3

#4 opened 12 days ago by

Jlfn

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 13 days ago

EXL3

❤️ 1

1

#2 opened 13 days ago by

XOHKEP

Infinite "hmm" loop in reasoning — is this patched yet?

2

#1 opened 13 days ago by

leonice2

New activity in dealignai/GLM-5.3-ABLITERATED-NVFP4 16 days ago

This model is broken with asian characters

5

#3 opened about 1 month ago by

nyxia

New activity in IFM/K2-Horizon-7B 22 days ago

I wish you delivered what you say you delivered

👍👀 5

17

#3 opened about 2 months ago by

NezTheNaughty

New activity in dealignai/Bonsai-2-27B-Ternary-CRACK-GGUF about 1 month ago

reasoning mode still looping

8

#1 opened about 1 month ago by

Jcamacho05

New activity in dealignai/DeepSeek-V4.1-Flash-UNCENSORED-FP8 about 2 months ago

Field report: UNCENSORED-FP8 serving at 1M context on 8x DGX Spark (GB10), vLLM TP8, with numbers

🔥 2

6

#1 opened about 2 months ago by

0xMagnus

New activity in dealignai/GLM-5.3-CYBERSECURITY-FP8 about 2 months ago

NVFP4

➕ 2

2

#2 opened about 2 months ago by

alonsoko

New activity in dealignai/GLM-5.3-UNCENSORED-FP8 about 2 months ago

Field report: FP8 v2 and NVFP4 on 8x DGX Spark (GB10) at TP=8 -- what works, what bites, and one template trap worth documenting

❤️ 1

9

#3 opened about 2 months ago by

0xMagnus

New activity in dealignai/Qwen3.8-Flash-Next-UNCENSORED-FP8 about 2 months ago

Reasoning-trace length: abliterated vs. base Qwen3.8-Flash-Next on GPQA Diamond

13

#2 opened about 2 months ago by

Soomin33

New activity in dealignai/Qwen3.8-Flash-Next-CYBERSECURITY-NVFP4 about 2 months ago

other formats

9

#2 opened about 2 months ago by

oneerrorx

New activity in dealignai/GLM-5.3-UNCENSORED-FP8 about 2 months ago

V2 still degenerates into word salad on real long-context agent workloads

2

#2 opened about 2 months ago by

leonice2

New activity in dealignai/Qwen3.8-Flash-Next-CYBERSECURITY-NVFP4 about 2 months ago

Reasoning improvements?

1

#1 opened about 2 months ago by

dealignai

New activity in dealignai/GLM-5.3-Flash-UNCENSORED-NVFP4 2 months ago

Token output random corruption, ModelOpt artifact?

1

#2 opened 2 months ago by

m-e-acc

Vision not working

3

#1 opened 2 months ago by

GulfCoastAI