GGUF models by piyawudk

36 GGUF models quantized by piyawudk on Hugging Face. Sizes run from 4.21B to 8.19B parameters. The longest trained context is 256K tokens.

Most downloadedNewestLargestLongest contextSmallest
QW
PhishMe 12k Qwen3.5 4B DPO v1
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe Qwen3 8B GRPO
Qwen · piyawudk
Params8.19B
Context40K
Q4_K_M5.0 GB
GGUF
GE
PhishMe 12k Gemma4 E4B SFT FFT
Gemma · piyawudk
Params4.65B
Context128K
Smallest5.0 GB
ReasoningTool useVision
GE
PhishMe 12k Gemma4 E4B SFT FFT GGUF v2
Gemma · piyawudk
Params4.65B
Context128K
Smallest5.0 GB
ReasoningTool useVision
QW
PhishMe 12k Qwen3.5 4B GSPO v9
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
PH
PhishMe R1 8B GRPO
Phishme · piyawudk
Params8.19B
Context128K
Q4_K_M5.0 GB
Reasoning
QW
PhishMe 12k Qwen3.5 4B SDFT v1
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P2 SDFT
Qwen · piyawudk
Params4.21B
Context10K
Smallest4.5 GB
ReasoningTool use
PH
PhishMe R1 8B SFT
Phishme · piyawudk
Params8.19B
Context128K
Q4_K_M5.0 GB
Reasoning
QW
PhishMe 12k Qwen3.5 4B P1 CSFT
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B GSPO v7
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CSFT v3
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe Qwen3 8B SFT
Qwen · piyawudk
Params8.19B
Context40K
Q4_K_M5.0 GB
GGUF
QW
PhishMe 12k Qwen3.5 4B GSPO v10
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P2 DPO
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CDFT v3
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B CorrectionSFT v1
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe Qwen3 Base 8B GRPO
Qwen · piyawudk
Params8.19B
Context32K
Q4_K_M5.0 GB
GGUF
QW
PhishMe 12k Qwen3.5 4B P1 CDFT v1
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe Qwen3 Base 8B SFT
Qwen · piyawudk
Params8.19B
Context32K
Q4_K_M5.0 GB
GGUF
QW
PhishMe 12k Qwen3.5 4B GSPO v8
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CSFT v1
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P2 DAPO Merged
Qwen · piyawudk
Params4.21B
Context8K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 DFT
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B FFT v6
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B FFT v5
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CDFT
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CDFT v2
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B CS
Qwen · piyawudk
Params4.21B
Context256K
Q4_K_M2.7 GB
ReasoningTool use
QW
PhishMe 12k Qwen3.5 4B P1 CSFT v2
Qwen · piyawudk
Params4.21B
Context256K
Smallest4.5 GB
ReasoningTool use

About piyawudk on Hugging Face

piyawudk has published 36 GGUF models that ggufmodels.com tracks, across families such as Qwen, Phishme and Gemma. The most downloaded are PhishMe 12k Qwen3.5 4B DPO v1, PhishMe Qwen3 8B GRPO and PhishMe 12k Gemma4 E4B SFT FFT.

The quants piyawudk publishes most often are Q8_0 (36), BF16 (28) and Q4_K_M (10). Their Q4_K_M files range from 2.7 to 5.0 GB. Parameter counts run from 4.21B to 8.19B, and the longest trained context window is 256K tokens.

Frequently asked questions

What GGUF models has piyawudk published?

36 models are listed here, mainly from the Qwen, Phishme and Gemma families. The most downloaded is PhishMe 12k Qwen3.5 4B DPO v1.

Which quants does piyawudk publish?

The most common are Q8_0 (36 models), BF16 (28 models) and Q4_K_M (10 models).

Keep browsing