Pratik Gandhi

pratikgx's activity

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

New activity in prism-ml/Ternary-Bonsai-2-27B-gguf about 2 hours ago

Independent test: 98% retention on easier tasks, 87% on harder reasoning/coding (vs 4-bit Qwen3.8 27B)

❤️ 2

4

#102 opened 1 day ago by

heldoutlab

New activity in RecursiveSI/flint about 2 hours ago

Memory needed per quant (file sizes + KV cache estimate)

3

#8 opened 3 days ago by

localllmrig

New activity in ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUF about 8 hours ago

Using IQ3_S-MTP on 9060XT(16) but got problems with overthinking

3

#59 opened 1 day ago by

HelloApplepie

New activity in RecursiveSI/flint about 8 hours ago

chat_template ignores enable_thinking=false, and reasoning leaks into content under vLLM

1

#7 opened 3 days ago by

DataCraftsmanAU

New activity in unsloth/DeepSeek-V4.1-Flash-1017-GGUF about 9 hours ago

Why so big vs unsloth/DeepSeek-V4-Flash-0731-GGUF

5

#2 opened 6 days ago by

dehnhaide

New activity in RecursiveSI/flint about 9 hours ago

Tool-use planning loops with rec despite attached tool schemas

2

#12 opened 2 days ago by

rachpradhan

New activity in dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8 about 9 hours ago

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

3

#6 opened 1 day ago by

wen-88-QQ

New activity in meta-models/Muse-Spark-Open-816B-A42B about 10 hours ago

Releasing Massive Open-Weight Models REALLY Matters

👍 5

5

#24 opened 7 days ago by

nono2221

New activity in mlx-community/Qwen3.8-27B-4bit about 11 hours ago

M6 Mac mini 24GB benchmark: context limits, prefill and MTP results

1

#4 opened 25 days ago by

lpalbou

New activity in prism-ml/Ternary-Bonsai-2-27B-gguf about 11 hours ago

Speculative Drafters for Ternary-Bonsai-2-27B (11~13 tok/s on Apple Silicon M3)

🔥 1

3

#100 opened 5 days ago by

NOhhh

New activity in Qwen/Qwen3.8-Flash-Next about 12 hours ago

How to limit thinking?

4

#71 opened 24 days ago by

MarbuelB

New activity in unsloth/Qwen3.8-27B-GGUF about 22 hours ago

Uniform GGUF quants silently break Qwen3.8-27B's deep thinking — reproduced on llama.cpp AND vLLM (short tasks unaffected)

11

#135 opened 19 days ago by

Jerryccccc

New activity in deepseek-ai/DeepSeek-V4.1-Flash-1017 about 23 hours ago

Did you know that you can run this model with 24 GB of VRAM and 64 GB of RAM?

🔥 7

14

#2 opened 8 days ago by

lucyferorg

New activity in unsloth/flint-GGUF about 23 hours ago

Does it work with llama.cpp?

❤️ 2

14

#1 opened 3 days ago by

auf1r2

New activity in RecursiveSI/flint 1 day ago

M5 Pro Mac mini (64GB) numbers

#15 opened 1 day ago by

pratikgx

New activity in mlx-community/Qwen3-ASR-0.6B-8bit 16 days ago

Add pipeline_tag and base_model metadata

#1 opened 16 days ago by

pratikgx

New activity in mlx-community/whisper-large-v3-turbo 16 days ago

Add license and base_model, fix source link and usage snippet

#6 opened 16 days ago by

pratikgx

New activity in mlx-community/gemma-4-26b-a4b-it-4bit 30 days ago

Add memory requirements, architecture notes, and usage guidance

#3 opened 30 days ago by

pratikgx

New activity in mlx-community/Qwen3.8-27B-8bit 30 days ago

Add memory requirements, architecture notes, and usage guidance

#2 opened 30 days ago by

pratikgx

New activity in mlx-community/Qwen3.8-27B-4bit 30 days ago

Add memory requirements, architecture notes, and usage guidance

#3 opened 30 days ago by

pratikgx