About 3.09B parameters
Safetensors metadata reports 3,085,938,688 parameters, or about 3.09B.
Open Source Model Profile · MegaScience
Qwen2.5-3B-MegaScience
Qwen2.5-3B-MegaScience is a 3.09B-parameter Qwen2 text-generation model from MegaScience. The model card describes it as one of the models trained in the MegaScience project on post-training datasets for science reasoning.
- Publisher
- MegaScience
- Task
- text-generation
- Model type
- qwen2
- License
- apache-2.0
- Library
- transformers
- Publication status
- Approved for indexing
Model overview
MegaScience publishes Qwen2.5-3B-MegaScience as a text-generation model. Captured config identifies Qwen2ForCausalLM with a qwen2 model type, and Safetensors metadata reports 3,085,938,688 parameters. Card data records apache-2.0. Hub tags list Qwen/Qwen2.5-3B as base model and finetune, and the captured model card's model-tree block names the same base.
Recorded capabilities
Apache-2.0 licensing
Captured metadata records an apache-2.0 license.
MegaScience project card
The model card describes this repository as a MegaScience project model trained on post-training datasets for science reasoning, and names MegaScience/MegaScience as the dataset used to train it.
Documented Transformers chat usage
The model card shows AutoTokenizer and AutoModelForCausalLM loading with apply_chat_template.
Use cases in the source record
- Transformers text-generation and chat-template workflows following the card's AutoModelForCausalLM example.
- Experiments described by the publisher as part of MegaScience post-training work on science-reasoning datasets.
Limitations and unknowns
- No evaluation scores were extracted; the card includes an Evaluation Results heading without captured metrics.
- Provider state is historical snapshot data, not independently refreshed current availability.
- Training recipe values, dataset assignment, and the Qwen/Qwen2.5-3B base-model listing come from hub tags and the publisher model card, not independently verified Ethen measurements. Max length 4,096 is a training-recipe field, not an independently verified context window.
Source and provenance
Source: MegaScience/Qwen2.5-3B-MegaScience
Captured: Unknown. Processed: 2026-09-07T19:35:41.739349+00:00.
MegaScience: Pushing the Frontiers of Post-Training Datasets for Science Reasoning This repository contains the Qwen2.5-3B-MegaScience model, one of the models trained as part of the MegaScience project. For the official code, data processing pipeline, and evaluation system, please refer to the MegaScience GitHub repository . Qwen2.5-3B-MegaScience Usage You can use this model with the Hugging Face transformers library: from transformers import AutoModelForCausalLM, AutoTokenizer model_name = "MegaScience/Qwen2.5-3B-MegaScience" tokenizer = AutoTokenizer.from_pretrained(model_name) model = AutoModelForCausalLM.from_pretrained(model_…
F001F002F003F004F005F006F007F009F010F011F014F015F016F017F018