Instructions to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3.5-27B")
model = PeftModel.from_pretrained(base_model, "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1")

Transformers

How to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 with Transformers:

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

pipe = pipeline("text-generation", model="praxisresearch/hf_qwen35_27b_sgtr_syspopped_1")
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, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("praxisresearch/hf_qwen35_27b_sgtr_syspopped_1")
model = AutoModelForMultimodalLM.from_pretrained("praxisresearch/hf_qwen35_27b_sgtr_syspopped_1", device_map="auto")
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 Settings

vLLM

How to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/praxisresearch/hf_qwen35_27b_sgtr_syspopped_1

SGLang

How to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 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 "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1" \
    --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": "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1",
		"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 "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1" \
        --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": "praxisresearch/hf_qwen35_27b_sgtr_syspopped_1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use praxisresearch/hf_qwen35_27b_sgtr_syspopped_1 with Docker Model Runner:
```
docker model run hf.co/praxisresearch/hf_qwen35_27b_sgtr_syspopped_1
```

See axolotl config

axolotl version: 0.18.0

adapter: lora

bf16: auto

- message_field_content: content
  message_field_role: role
  path: data/finetuning/sgtr/detection/prefer-self-finetune_target_hf_qwen35_27b_other-models__claude-21__finetuningdata_syspoppedprompt.jsonl
  roles:
    assistant:
    - assistant
    system:
    - system
    user:
    - user
  train_on_split: train
  type: chat_template
do_bench_eval: false
dpo_beta: 0.1
eval_batch_size: null
eval_sample_packing: false
eval_steps: null
fp16: false
gradient_accumulation_steps: 8
gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
group_by_length: false
learning_rate: 1.0e-05
logging_steps: 1
lora_alpha: 64
lora_dropout: 0.0
lora_fan_in_fan_out: false
lora_mlp_kernel: false
lora_model_dir: null
lora_o_kernel: false
lora_qkv_kernel: false
lora_r: 32
lora_target_modules:
- q_proj
- k_proj
- v_proj
- o_proj
- in_proj_qkv
- in_proj_a
- in_proj_b
- in_proj_z
- out_proj
- gate_proj
- up_proj
- down_proj
lr_scheduler: linear
micro_batch_size: 2
model_type: AutoModelForCausalLM
num_epochs: 1
optimizer: adamw_8bit
output_dir: models/hf_qwen35_27b_sgtr_syspopped_1
pad_to_sequence_len: false
peft_use_dora: false
peft_use_rslora: true
push_to_hub: false
save_safetensors: true
saves_per_epoch: 1
seed: 1
sequence_len: 2048
special_tokens: null
strict: false
tf32: false
tokenizer_type: AutoTokenizer
train_on_inputs: false
val_set_size: 0
wandb_log_model: null
wandb_project: hf_qwen35_27b_sgtr_syspopped_1
wandb_run_id: null
wandb_watch: null
warmup_steps: 5
weight_decay: 0.01

models/hf_qwen35_27b_sgtr_syspopped_1

This model is a fine-tuned version of Qwen/Qwen3.5-27B on the data/finetuning/sgtr/detection/prefer-self-finetune_target_hf_qwen35_27b_other-models__claude-21__finetuningdata_syspoppedprompt.jsonl dataset.

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-05
train_batch_size: 2
eval_batch_size: 2
seed: 1
gradient_accumulation_steps: 8
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 5
training_steps: 125

Training results

Framework versions

PEFT 0.19.1
Transformers 5.3.0.dev0
Pytorch 2.11.0+cu128
Datasets 4.8.4
Tokenizers 0.22.2

Provenance

sgtr_syspopped training applied directly to the base model (stage 1 of the prevention arm).

Base model: Qwen/Qwen3.5-27B
Seed: 1
Training chain: Qwen/Qwen3.5-27B -> sgtr_syspopped
Stage data: per-seed self-recognition (SGTR) detection data, system message removed
Method: LoRA (r=32, alpha=64, rsLoRA), 1 epoch, lr 1e-5, seq len 2048. Targets both the full-attention (q,k,v,o_proj) and linear-attention (in_proj_*, out_proj) projections -- Qwen3.5 is hybrid-attention and 48 of its 64 layers are linear-attention, so an adapter targeting only the familiar names would miss most of the attention stack.

How to use

Load directly on top of Qwen/Qwen3.5-27B.

Note on precision

Evaluate in bfloat16 (the checkpoint dtype). Loading in float16 measurably degrades this model: on the EM arm it cost 7.3 points of TruthfulQA accuracy.

Part of the SGTR/EM research project: http://tiny.cc/llm_self_recognition

Downloads last month: 10