From the model card
Fields below are copied from the tags and counters on the HuggingFace repository Qwen/Qwen2.5-1.5B-Instruct at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- Qwen
- Pipeline tag
- text-generation
- Library
- Transformers
- Weight formats
- safetensors
- License tag
apache-2.0— read the license file in the repo before relying on it- Lineage
-
- base model Qwen/Qwen2.5-1.5B
- fine-tune of Qwen/Qwen2.5-1.5B
- Language tags
- English (en)
- Papers cited
- arXiv:2407.10671
- Downloads (HF counter at last fetch)
- 7,383,027
- Likes (HF counter at last fetch)
- 816
- Model card
- https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct
Use cases
- Embedded on-device inference on constrained hardware
- Simple instruction following tasks like classification, reformatting, or short summarization
- Ultra-low-latency text generation where quality is secondary to speed
- Prototyping LLM features with minimal infrastructure
- Lightweight chat on CPU-only servers
Pros
- Apache 2.0 license
- 1.5B parameters runs on very limited hardware including CPU
- Part of maintained Qwen2.5 family
- Text-generation-inference compatible
Cons
- 1.5B scale significantly limits reasoning, factual accuracy, and coherent multi-turn dialogue
- Not competitive with 3B+ models on most benchmarks
- Hallucination rate high relative to larger models
- Complex tasks requiring multi-step reasoning are unreliable
- Context window and multilingual breadth more limited than larger family members
Tags
transformerssafetensorsqwen2text-generationchatconversationalenarxiv:2407.10671base_model:Qwen/Qwen2.5-1.5Bbase_model:finetune:Qwen/Qwen2.5-1.5Blicense:apache-2.0text-generation-inferenceendpoints_compatibleregion:usdeploy:azure