Instructions to use DreamFast/Qwen-2.5-7b-apostate with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DreamFast/Qwen-2.5-7b-apostate with Transformers:

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

pipe = pipeline("text-generation", model="DreamFast/Qwen-2.5-7b-apostate")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DreamFast/Qwen-2.5-7b-apostate")
model = AutoModelForCausalLM.from_pretrained("DreamFast/Qwen-2.5-7b-apostate", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use DreamFast/Qwen-2.5-7b-apostate with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DreamFast/Qwen-2.5-7b-apostate"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DreamFast/Qwen-2.5-7b-apostate",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DreamFast/Qwen-2.5-7b-apostate

SGLang

How to use DreamFast/Qwen-2.5-7b-apostate 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 "DreamFast/Qwen-2.5-7b-apostate" \
    --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": "DreamFast/Qwen-2.5-7b-apostate",
		"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 "DreamFast/Qwen-2.5-7b-apostate" \
        --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": "DreamFast/Qwen-2.5-7b-apostate",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DreamFast/Qwen-2.5-7b-apostate with Docker Model Runner:
```
docker model run hf.co/DreamFast/Qwen-2.5-7b-apostate
```

Qwen-2.5-7b-apostate / report.json

DreamFast

Upload folder using huggingface_hub

afee64c verified 2 months ago

Raw

History Blame Contribute Delete

2.55 kB

	{
	"model": "Qwen/Qwen2.5-7B-Instruct",
	"num_layers": 28,
	"hidden_size": 3584,
	"direction_layer": 20,
	"refusal_subspace_rank": 1,
	"initial_separation": 74.8803,
	"baseline_refusal_rate": 0.9583,
	"baseline_eval_n": 24,
	"edited_refusal_rate": 0.2286,
	"refusal_metric": "classifier + weak guard",
	"harmless_kl_nats": 0.0903,
	"kl_backoff_steps": 0,
	"kl_layer_trim_steps": 0,
	"repair_steps": 1,
	"guard_history": [
	{
	"iter": 0,
	"separation": 40.6304,
	"ratio": 0.5426,
	"rank": 1,
	"refusal": 0.75,
	"kl": 0.0128
	},
	{
	"iter": 1,
	"separation": 40.7273,
	"ratio": 0.5439,
	"rank": 1,
	"refusal": 0.4688,
	"kl": 0.0278
	}
	],
	"layer_alphas": [
	0.667,
	0.667,
	0.667,
	0.667,
	0.667,
	0.667,
	1.206,
	1.222,
	1.205,
	1.202,
	1.21,
	0.0,
	1.24,
	1.333,
	1.333,
	1.333,
	1.333,
	1.202,
	1.448,
	1.333,
	1.333,
	0.667,
	0.667,
	0.667,
	0.667,
	0.667,
	0.667,
	0.667
	],
	"embed_alpha": 0.059,
	"head_alpha": 0.0,
	"preserve_rank": 8,
	"preserve_source": "harmless",
	"pruned_layers": [],
	"layers_after_prune": 28,
	"elapsed_sec": 271.2,
	"profile": "balanced",
	"target_refusal": 0.03,
	"max_kl": 0.16,
	"kl_target": 0.06,
	"kl_positions": 32,
	"opt_capability": true,
	"opt_capability_weight": 1.0,
	"timings_sec": {
	"load_model": 12.5,
	"load_prompts": 16.0,
	"baseline_refusal": 5.2,
	"activation_fit": 13.7,
	"causal_scores": 12.2,
	"optimize_profile": 90.7,
	"guard": 22.4,
	"refine_refusal": 14.3,
	"validation_metrics": 0.0,
	"repair": 57.8,
	"prune": 0.0,
	"test_metrics": 13.6,
	"bake": 13.0
	},
	"command": "apostate ablate --model Qwen/Qwen2.5-7B-Instruct --out qwen-apostate",
	"optimized": true,
	"best_params": {
	"direction_source": "activations",
	"direction_layer_frac": 0.7420306856024312,
	"refusal_rank": 1,
	"strength": 1.0860947545472213,
	"band_center": 0.48999396234759546,
	"band_width": 0.5636482354840956,
	"causal_mix": 0.664168644393967,
	"causal_power": 1.2972323636074856,
	"direction_sign": 1.0,
	"ablate_embed": true,
	"embed_scale": 0.04047498432630894,
	"ablate_head": false,
	"head_scale": 0.010282217886773384,
	"head_alpha": 0.6950566529600242
	},
	"best_trial": {
	"refusal": 0.75,
	"kl": 0.0128,
	"capability_logprob": -7.9867,
	"capability_drift": 0.0
	},
	"n_trials": 16,
	"baked_to": "qwen-apostate"
	}