Modelpublic

theprint/GameMaster-v1-2B-GGUF

sourceHugging Faceupdated 2mo agoView on Hugging Face

0likes792downloads

Model Card

GameMaster-v1-2B (GGUF)

A fine-tuned version of `unsloth/Qwen3.5-2B` trained on GameMastering sharegpt data using Auto-SFT — an automated hyperparameter search and supervised fine-tuning pipeline.

The base model was adapted to follow the style and content of the GameMastering sharegpt dataset. Expect improved performance on tasks similar to those represented in the training data.

Model Details

Property	Value
Base model	`unsloth/Qwen3.5-2B`
Training data	`data/GameMastering-sharegpt.json`
Fine-tuning epochs	2
Fine-tuning date	2026-07-19
Fine-tuning method	LoRA (merged to full 16-bit)

Training Hyperparameters

LoRA

Parameter	Value
`r`	`4`
`alpha`	`16`
`dropout`	`0.04`
`target_modules`	`['q_proj', 'v_proj', 'k_proj', 'o_proj']`

Training

Parameter	Value
`learning_rate`	`1e-05`
`batch_size`	`4`
`gradient_accumulation_steps`	`8`
`warmup_ratio`	`0.03`
`max_seq_length`	`2048`
`quantization`	`none`

GGUF Files

These quantized GGUF files can be used directly with llama.cpp, Ollama, LM Studio, and other compatible runtimes.

File	Description
`GameMaster-v1-2B-GGUF-BF16.gguf`	BF16
`GameMaster-v1-2B-GGUF-Q8_0.gguf`	8-bit — near-lossless, larger file
`GameMaster-v1-2B-GGUF-Q6_K.gguf`	6-bit — high quality
`GameMaster-v1-2B-GGUF-Q5_K_M.gguf`	5-bit medium — good quality/size balance
`GameMaster-v1-2B-GGUF-Q5_K_S.gguf`	Q5KS
`GameMaster-v1-2B-GGUF-Q4_K_M.gguf`	4-bit medium — recommended for most use cases
`GameMaster-v1-2B-GGUF-Q4_K_S.gguf`	Q4KS
`GameMaster-v1-2B-GGUF-Q3_K_L.gguf`	Q3KL
`GameMaster-v1-2B-GGUF-Q3_K_M.gguf`	Q3KM
`GameMaster-v1-2B-GGUF-Q3_K_S.gguf`	Q3KS
`GameMaster-v1-2B-GGUF-Q2_K.gguf`	2-bit — smallest size, lowest quality
`GameMaster-v1-2B-GGUF-IQ4_XS.gguf`	IQ4_XS
`GameMaster-v1-2B-GGUF-IQ4_NL.gguf`	IQ4_NL
`GameMaster-v1-2B-GGUF-TQ2_0.gguf`	TQ2_0

Generated by [Auto-SFT](https://github.com/theprint/auto-sft)