Instructions to use ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2")
model = AutoModelForImageTextToText.from_pretrained("ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2

SGLang

How to use ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2 with Docker Model Runner:
```
docker model run hf.co/ToastyPigeon/Qwen3.5-27B-Marvin-DPO-V2
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Qwen3.5-27B-Marvin-DPO-V2

A Qwen3.5-27B model fine-tuned for high-quality creative writing and roleplay, with DPO applied to reduce repetition, suppress AI-isms, and improve writing style.

Model Stack

Qwen/Qwen3.5-27B
  → ArliAI/Qwen3.5-27B-Derestricted (safety filter removal)
    → SFT: 5,974 samples (4,478 Marvin literary + 1,497 Seed RP)
      = ToastyPigeon/Qwen3.5-27B-Marvin-V2
        → DPO: 402 combined preference pairs
          = This model (Marvin-DPO-V2)

DPO Training Details

Combined DPO with three objectives trained simultaneously in a single run:

Subset	Pairs	Purpose
Anti-repetition (rewritten)	102	Suppress sentence/paragraph-level repetition. Thinking traces rewritten from verbose (avg 704w) to concise (avg 91w).
Anti-repetition (RP context)	100	Anti-repetition in roleplay scenarios. 20 unique character/setting combinations.
Style cleanup	200	Improve prose quality. Chosen: Marvin literary corpus excerpts. Rejected: model-generated versions of the same scenes. 50% book-style / 30% asterisk-action / 20% mixed format.

Think masking: DPO loss is computed only on the response content after </think>, not on the thinking traces themselves. This prevents the DPO signal from accidentally training away the model's ability to think.

Hyperparameters

DPO beta: 0.1
Loss type: Sigmoid
Learning rate: 5e-6 (cosine schedule, 10% warmup)
LoRA: r=32, alpha=16, RSLoRA, no dropout
Quantization: QLoRA (NF4)
Precision: bf16
Batch size: 1 × 4 grad accumulation = effective 4
Epochs: 1
Training time: ~68 minutes on 2× RTX 3090

Training Metrics

Train loss: 0.117 avg
Reward accuracy: 100%
Reward margins: 6-8 (strong chosen/rejected separation)

Evaluation

Tested across 5 scenarios (temp=0.8, top_p=0.9):

Test	-ing patterns	Notes
RP coffeeshop scene	0	Natural dialogue, good pacing
Hemingway style transfer	1	Short declarative sentences, understated
Chandler noir style	4	Vivid metaphors, atmospheric
Emotional scene (slop trap)	2	Grounded, no AI-isms
Instruction following	1	Doesn't write for user

Anti-Repetition (8-turn multi-turn test)

Model	Repeated 4/5-grams (3+)	Total -ing
V2 Base (no DPO)	6× "corner of my", 4× "tugging at the", 3× "loose strand"	4
DPO-V2 (this model)	1× "corner of her" at 3×	9

Zero sentence-level repetition across 8 turns of conversation, compared to significant repetition in the base model by turn 6.

Recommended Settings

Temperature: 0.8
Top-p: 0.9
Format: "Quotation marks" for speech, plain text for narration, *italics* for inner thoughts

Limitations

Ethiopian Yirgacheffe appears disproportionately when the model discusses coffee (baked into base model training data)
Thinking mode is suppressed — the model produces empty think blocks. Use <think>\n\n</think>\n\n prefill for non-thinking mode.
Participial phrase patterns (-ing) are reduced but not eliminated

Training Config

train-v2.yaml (click to expand)

# Combined DPO V2: antirep + style + thinking — on Marvin V2 base
# 402 pairs, 1 epoch, beta=0.1, LR=5e-6
# V2: think masking enabled, 63% of pairs have think blocks

model_name_or_path: ToastyPigeon/Qwen3.5-27B-Marvin-V2
output_dir: runs/qwen35-27b-combined-dpo-v2

attn_implementation: flash_attention_2
bf16: true
gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false

model_parallel: true
max_memory:
  0: "18GiB"
  1: "18GiB"

chunked_mlp: true
chunked_mlp_chunks: 8

max_length: 2048
max_prompt_length: 512
max_completion_length: 1536

use_chunked_dpo: true
chunked_dpo_size: 4096
precompute_ref_log_probs: true
mask_thinking: true

per_device_train_batch_size: 1
gradient_accumulation_steps: 4

use_peft: true
load_in_4bit: true
bnb_4bit_quant_type: nf4
lora_r: 32
lora_alpha: 16
lora_dropout: 0.0
use_rslora: true
lora_target_modules:
  - in_proj_qkv
  - in_proj_z
  - in_proj_a
  - in_proj_b
  - out_proj
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - gate_proj
  - up_proj
  - down_proj

beta: 0.1
loss_type: sigmoid

learning_rate: 5.0e-6
lr_scheduler_type: cosine
warmup_ratio: 0.1
weight_decay: 0.0
max_grad_norm: 1.0
optim: paged_adamw_8bit
num_train_epochs: 1

logging_steps: 1
save_strategy: epoch
save_total_limit: 1
report_to: none