---
license: apache-2.0
datasets:
- AI-MO/NuminaMath-CoT
- prithivMLmods/Math-Solve
- amphora/QwQ-LongCoT-130K
- prithivMLmods/Deepthink-Reasoning
- NovaSky-AI/Sky-T1_data_17k
language:
- en
base_model:
- deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
pipeline_tag: text-generation
library_name: transformers
tags:
- QwQ
- Distill
- R1
- Deepseek
- Qwen2.5
- text-generation-inference
---

# **QWQ R1 [Reasoning] Distill 1.5B CoT**

QWQ R1 [Reasoning] Distill 1.5B CoT is a fine-tuned language model designed for advanced reasoning and instruction-following tasks. It leverages the Qwen2.5 R1 Distill from the DeepSeek base model and has been fine-tuned on chain-of-thought (CoT) reasoning datasets, focusing on CoT reasoning for problem-solving. This model is optimized for tasks requiring logical reasoning, detailed explanations, and multi-step problem-solving, making it ideal for applications such as instruction-following, text generation, and complex reasoning tasks.

# **Quickstart with Transformers**

Here provides a code snippet with `apply_chat_template` to show you how to load the tokenizer and model and how to generate contents.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "prithivMLmods/QwQ-R1-Distill-1.5B-CoT"

model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)

prompt = "How many r in strawberry."
messages = [
    {"role": "system", "content": "You are a helpful and harmless assistant. You are Qwen developed by Alibaba. You should think step-by-step."},
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=512
)
generated_ids = [
    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]

response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
```
# **Intended Use**

**QWQ R1 [Reasoning] Distill 1.5B CoT** is specifically designed for tasks requiring advanced reasoning, structured thinking, and detailed explanations. Its intended applications include:

1. **Instruction-Following Tasks**: Performing step-by-step tasks based on user instructions.
2. **Logical Reasoning**: Solving problems that demand multi-step logical processing and inference.
3. **Text Generation**: Crafting coherent and contextually appropriate text for various domains.
4. **Educational Tools**: Assisting in learning environments, providing explanations for complex topics, or guiding through reasoning exercises.
5. **Problem-Solving**: Addressing computational or real-world problems requiring chain-of-thought reasoning.
6. **AI-Assisted Decision-Making**: Supporting users in making informed decisions with logical analysis.

# **Limitations**

While the model excels in reasoning and explanation tasks, it has certain constraints:

1. **Context Length**: Limited ability to process or generate outputs for inputs exceeding its maximum token limit.
2. **Domain Knowledge**: It may lack detailed expertise in niche domains not covered during training.
3. **Dependence on Training Data**: Performance can be influenced by biases or gaps in the datasets it was fine-tuned on.
4. **Real-Time Reasoning**: Struggles with tasks requiring dynamic understanding of real-time data or rapidly changing contexts.
5. **Mathematical Precision**: May produce errors in calculations or fail to interpret ambiguous mathematical problems.
6. **Factual Accuracy**: Occasionally generates incorrect or outdated information when dealing with facts.
7. **Language Nuances**: Subtle linguistic or cultural nuances might be misunderstood or misrepresented.
8. **Complex CoT Chains**: For extremely lengthy or convoluted reasoning chains, the model may lose track of earlier context or steps.