Instructions to use zbeeb/Qwen-2.5-7B-withVector with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zbeeb/Qwen-2.5-7B-withVector with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("zbeeb/Qwen-2.5-7B-withVector")
model = AutoModelForCausalLM.from_pretrained("zbeeb/Qwen-2.5-7B-withVector", 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 zbeeb/Qwen-2.5-7B-withVector with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/zbeeb/Qwen-2.5-7B-withVector

SGLang

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

Docker Model Runner
How to use zbeeb/Qwen-2.5-7B-withVector with Docker Model Runner:
```
docker model run hf.co/zbeeb/Qwen-2.5-7B-withVector
```

Improve model card: Add pipeline tag, paper, project page, and code links

by nielsr HF Staff - opened Sep 4, 2025

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+10

-4

Files changed (1) hide show

README.md +10 -4

README.md CHANGED Viewed

@@ -4,11 +4,18 @@ library_name: transformers
 tags:
 - mergekit
 - merge
 ---
 # mergercoder
-This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
 ## Merge Details
 ### Merge Method
@@ -36,5 +43,4 @@ models:
       weight: -1.0
 merge_method: task_arithmetic
 dtype: bfloat16
-```
- arxiv.org/abs/2509.01363

 tags:
 - mergekit
 - merge
+pipeline_tag: text-generation
 ---
 # mergercoder
+This repository hosts a merged language model, based on the work presented in [Reasoning Vectors: Transferring Chain-of-Thought Capabilities via Task Arithmetic](https://arxiv.org/abs/2509.01363). This work demonstrates that reasoning ability, once learned, can be extracted and transferred between models as a compact task vector, offering a practical way to enhance models by recycling prior computational investments.
+*   📖 Paper: [Reasoning Vectors: Transferring Chain-of-Thought Capabilities via Task Arithmetic](https://arxiv.org/abs/2509.01363)
+*   🌐 Project Page: [https://elm.baulab.info](https://elm.baulab.info)
+*   💻 Code: [https://github.com/rohitgandikota/erasing-llm](https://github.com/rohitgandikota/erasing-llm)
+This model is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
 ## Merge Details
 ### Merge Method
       weight: -1.0
 merge_method: task_arithmetic
 dtype: bfloat16
+```