🤝 Open to Collab

Yinan Long

Brooooooklyn

3 1 7

sypher9's profile picture

Revangar's profile picture

goal2026hao's profile picture

https://github.com/Brooooooklyn

Brooooook_lyn
Brooooooklyn

AI & ML interests

None yet

Recent Activity

reacted to danielhanchen's post with 🔥 about 12 hours ago

Introducing Unsloth for AMD 🚀 You can now train & run LLMs on your AMD hardware • We collaborated with AMD to enable you to train & run 500+ models on AMD GPUs • Works on Windows, WSL, Linux • Train Qwen, Gemma on just 3GB VRAM GitHub: https://github.com/unslothai/unsloth Blog + Guide: https://unsloth.ai/docs/basics/amd

updated a collection 3 days ago

Qwen Unsloth NVFP4 Tensor-Class Recipe for MLX — macOS + DGX

updated a collection 3 days ago

Qwen Unsloth NVFP4 Tensor-Class Recipe for MLX — macOS + DGX

View all activity

Organizations

None yet

Brooooooklyn 's collections 8

Qwen Unsloth NVFP4 Tensor-Class Recipe for MLX — macOS + DGX

Unsloth Qwen3.6 tensor-class map for MLX: MXFP4+MXFP8 on macOS; NVFP4+E4M3 weights with A16 on DGX.

Brooooooklyn/Agents-A1-mxfp4-mlx

Text Generation • 8B • Updated 3 days ago • 38
Brooooooklyn/Agents-A1-nvfp4-mlx

Text Generation • 24B • Updated 3 days ago • 50
Brooooooklyn/Ornith-1.0-35B-mxfp4-mlx

Text Generation • 7B • Updated 3 days ago • 760
Brooooooklyn/Ornith-1.0-35B-nvfp4-mlx

Text Generation • 24B • Updated 3 days ago • 824

Qwen-AgentWorld-35B-A3B-unsloth-mlx

Qwen-AgentWorld-35B-A3B (Qwen3.5-VL-MoE world model) for Apple Silicon MLX: legacy imatrix-AWQ Q3–Q8 + MXFP8, quality-verified vs BF16.

Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q3_K_XL-mlx

Image-Text-to-Text • 5B • Updated 21 days ago • 249 • 1
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q4_K_XL-mlx

Image-Text-to-Text • 6B • Updated 21 days ago • 351
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q5_K_XL-mlx

Image-Text-to-Text • 7B • Updated 21 days ago • 300
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q6_K_XL-mlx

Image-Text-to-Text • 8B • Updated 21 days ago • 333

Gemma-4-31B-IT-unsloth-mlx

Gemma-4-31B-IT (dense vision-language) quantized for Apple Silicon (MLX) — Unsloth Dynamic 2.0 with AWQ imatrix pre-scaling.

Brooooooklyn/Gemma-4-31B-IT-UD-Q2_K_XL-mlx

Text Generation • 7B • Updated May 16 • 34
Brooooooklyn/Gemma-4-31B-IT-UD-Q3_K_XL-mlx

Text Generation • 8B • Updated May 16 • 31
Brooooooklyn/Gemma-4-31B-IT-UD-Q4_K_XL-mlx

Text Generation • 9B • Updated May 16 • 108
Brooooooklyn/Gemma-4-31B-IT-UD-MXFP4_K_XL-mlx

Text Generation • 9B • Updated May 16 • 19 • 1

Qwen-3.6-unsloth-mlx

AWQ-style pre-scaling using Unsloth's imatrix calibration data, then 3-6-bit affine quantization with the Unsloth mixed-precision recipe via MLX

Brooooooklyn/Qwen3.6-35B-A3B-UD-Q2_K_XL-mlx

Text Generation • 4B • Updated May 15 • 651 • 7
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q3_K_XL-mlx

Text Generation • 6B • Updated May 15 • 364 • 4
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q4_K_XL-mlx

Text Generation • 7B • Updated May 15 • 1.42k • 9
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q5_K_XL-mlx

Text Generation • 8B • Updated May 15 • 252 • 2

NVIDIA NVFP4 → MLX

NVIDIA post-trained NVFP4 checkpoints (ModelOpt PTQ) transcoded to MLX for Apple Silicon / mlx-node, keeping NVIDIA's per-layer bit allocation.

Brooooooklyn/Gemma-4-26B-A4B-NVFP4-mlx

Text Generation • 9B • Updated 18 days ago • 442 • 1

Ornith-1.0-35B-unsloth-mlx

Ornith-1.0-35B for Apple Silicon MLX: legacy Unsloth Dynamic Q3–Q8 + MXFP8. Output quality verified vs BF16.

Brooooooklyn/Ornith-1.0-35B-UD-Q3_K_XL-mlx

Text Generation • 5B • Updated 21 days ago • 633
Brooooooklyn/Ornith-1.0-35B-UD-Q4_K_XL-mlx

Text Generation • 6B • Updated 21 days ago • 608
Brooooooklyn/Ornith-1.0-35B-UD-Q5_K_XL-mlx

Text Generation • 8B • Updated 21 days ago • 395
Brooooooklyn/Ornith-1.0-35B-UD-Q6_K_XL-mlx

Text Generation • 9B • Updated 21 days ago • 823 • 1

Gemma-4-26B-IT-unsloth-mlx

Gemma-4-26B-A4B-IT MoE quantized for Apple Silicon (MLX) — Unsloth Dynamic 2.0 with AWQ imatrix pre-scaling.

Brooooooklyn/Gemma-4-26B-A4B-IT-UD-Q3_K_XL-mlx

Text Generation • 4B • Updated May 15 • 128
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-MXFP4_K_XL-mlx

Text Generation • 6B • Updated May 15 • 52 • 1
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-Q4_K_XL-mlx

Text Generation • 5B • Updated May 15 • 169
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-NVFP4_K_XL-mlx

Text Generation • 7B • Updated May 15 • 23

Qwen-3.5-unsloth-mlx

AWQ-style pre-scaling using Unsloth's imatrix calibration data, then 3-6-bit affine quantization with the Unsloth mixed-precision recipe via MLX

Brooooooklyn/Qwen3.5-9B-unsloth-mlx

Text Generation • 2B • Updated Mar 24 • 673 • 12
Brooooooklyn/Qwen3.5-27B-unsloth-mlx

Text Generation • 6B • Updated Mar 24 • 761 • 18
Brooooooklyn/Qwen3.5-35B-A3B-UD-Q2_K_XL-mlx

Text Generation • 4B • Updated Mar 29 • 41
Brooooooklyn/Qwen3.5-35B-A3B-UD-Q3_K_XL-mlx

Text Generation • 6B • Updated Mar 29 • 43 • 2

Qwen Unsloth NVFP4 Tensor-Class Recipe for MLX — macOS + DGX

Unsloth Qwen3.6 tensor-class map for MLX: MXFP4+MXFP8 on macOS; NVFP4+E4M3 weights with A16 on DGX.

Brooooooklyn/Agents-A1-mxfp4-mlx

Text Generation • 8B • Updated 3 days ago • 38
Brooooooklyn/Agents-A1-nvfp4-mlx

Text Generation • 24B • Updated 3 days ago • 50
Brooooooklyn/Ornith-1.0-35B-mxfp4-mlx

Text Generation • 7B • Updated 3 days ago • 760
Brooooooklyn/Ornith-1.0-35B-nvfp4-mlx

Text Generation • 24B • Updated 3 days ago • 824

NVIDIA NVFP4 → MLX

NVIDIA post-trained NVFP4 checkpoints (ModelOpt PTQ) transcoded to MLX for Apple Silicon / mlx-node, keeping NVIDIA's per-layer bit allocation.

Brooooooklyn/Gemma-4-26B-A4B-NVFP4-mlx

Text Generation • 9B • Updated 18 days ago • 442 • 1

Qwen-AgentWorld-35B-A3B-unsloth-mlx

Qwen-AgentWorld-35B-A3B (Qwen3.5-VL-MoE world model) for Apple Silicon MLX: legacy imatrix-AWQ Q3–Q8 + MXFP8, quality-verified vs BF16.

Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q3_K_XL-mlx

Image-Text-to-Text • 5B • Updated 21 days ago • 249 • 1
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q4_K_XL-mlx

Image-Text-to-Text • 6B • Updated 21 days ago • 351
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q5_K_XL-mlx

Image-Text-to-Text • 7B • Updated 21 days ago • 300
Brooooooklyn/Qwen-AgentWorld-35B-A3B-UD-Q6_K_XL-mlx

Image-Text-to-Text • 8B • Updated 21 days ago • 333

Ornith-1.0-35B-unsloth-mlx

Ornith-1.0-35B for Apple Silicon MLX: legacy Unsloth Dynamic Q3–Q8 + MXFP8. Output quality verified vs BF16.

Brooooooklyn/Ornith-1.0-35B-UD-Q3_K_XL-mlx

Text Generation • 5B • Updated 21 days ago • 633
Brooooooklyn/Ornith-1.0-35B-UD-Q4_K_XL-mlx

Text Generation • 6B • Updated 21 days ago • 608
Brooooooklyn/Ornith-1.0-35B-UD-Q5_K_XL-mlx

Text Generation • 8B • Updated 21 days ago • 395
Brooooooklyn/Ornith-1.0-35B-UD-Q6_K_XL-mlx

Text Generation • 9B • Updated 21 days ago • 823 • 1

Gemma-4-31B-IT-unsloth-mlx

Gemma-4-31B-IT (dense vision-language) quantized for Apple Silicon (MLX) — Unsloth Dynamic 2.0 with AWQ imatrix pre-scaling.

Brooooooklyn/Gemma-4-31B-IT-UD-Q2_K_XL-mlx

Text Generation • 7B • Updated May 16 • 34
Brooooooklyn/Gemma-4-31B-IT-UD-Q3_K_XL-mlx

Text Generation • 8B • Updated May 16 • 31
Brooooooklyn/Gemma-4-31B-IT-UD-Q4_K_XL-mlx

Text Generation • 9B • Updated May 16 • 108
Brooooooklyn/Gemma-4-31B-IT-UD-MXFP4_K_XL-mlx

Text Generation • 9B • Updated May 16 • 19 • 1

Gemma-4-26B-IT-unsloth-mlx

Gemma-4-26B-A4B-IT MoE quantized for Apple Silicon (MLX) — Unsloth Dynamic 2.0 with AWQ imatrix pre-scaling.

Brooooooklyn/Gemma-4-26B-A4B-IT-UD-Q3_K_XL-mlx

Text Generation • 4B • Updated May 15 • 128
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-MXFP4_K_XL-mlx

Text Generation • 6B • Updated May 15 • 52 • 1
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-Q4_K_XL-mlx

Text Generation • 5B • Updated May 15 • 169
Brooooooklyn/Gemma-4-26B-A4B-IT-UD-NVFP4_K_XL-mlx

Text Generation • 7B • Updated May 15 • 23

Qwen-3.6-unsloth-mlx

AWQ-style pre-scaling using Unsloth's imatrix calibration data, then 3-6-bit affine quantization with the Unsloth mixed-precision recipe via MLX

Brooooooklyn/Qwen3.6-35B-A3B-UD-Q2_K_XL-mlx

Text Generation • 4B • Updated May 15 • 651 • 7
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q3_K_XL-mlx

Text Generation • 6B • Updated May 15 • 364 • 4
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q4_K_XL-mlx

Text Generation • 7B • Updated May 15 • 1.42k • 9
Brooooooklyn/Qwen3.6-35B-A3B-UD-Q5_K_XL-mlx

Text Generation • 8B • Updated May 15 • 252 • 2

Qwen-3.5-unsloth-mlx

AWQ-style pre-scaling using Unsloth's imatrix calibration data, then 3-6-bit affine quantization with the Unsloth mixed-precision recipe via MLX

Brooooooklyn/Qwen3.5-9B-unsloth-mlx

Text Generation • 2B • Updated Mar 24 • 673 • 12
Brooooooklyn/Qwen3.5-27B-unsloth-mlx

Text Generation • 6B • Updated Mar 24 • 761 • 18
Brooooooklyn/Qwen3.5-35B-A3B-UD-Q2_K_XL-mlx

Text Generation • 4B • Updated Mar 29 • 41
Brooooooklyn/Qwen3.5-35B-A3B-UD-Q3_K_XL-mlx

Text Generation • 6B • Updated Mar 29 • 43 • 2