baban/QwenTranslate_English_Hindi_100K_SFT
TEXT GENERATIONConcurrent Unit Cost:1Model Size:3.1BQuant:BF16Context Size:32kTool Calling:SupportedPublished:Jun 12, 2026License:otherArchitecture:Transformer Featherless Exclusive Cold
The baban/QwenTranslate_English_Hindi_100K_SFT model is a fine-tuned version of the Qwen/Qwen2.5-3B-Instruct architecture, specifically optimized for English-Hindi translation tasks. This model leverages the base Qwen2.5-3B-Instruct model and has been fine-tuned on the MT_En_Hindi dataset. It is designed to provide translation capabilities between English and Hindi, demonstrating a validation loss of 0.5699.
Loading preview...
Overview
This model, baban/QwenTranslate_English_Hindi_100K_SFT, is a specialized fine-tuned variant of the Qwen/Qwen2.5-3B-Instruct base model. Its primary purpose is to facilitate English-Hindi translation, having been trained on the MT_En_Hindi dataset.
Key Capabilities
- English-Hindi Translation: Optimized for translating text between English and Hindi.
- Fine-tuned Performance: Achieved a validation loss of 0.5699 during training, indicating its proficiency in the target task.
Training Details
The model underwent training with specific hyperparameters:
- Learning Rate: 5e-05
- Batch Size: A total training batch size of 1024 (8 per device across 8 GPUs with 16 gradient accumulation steps).
- Optimizer: ADAMW_TORCH.
- Epochs: Trained for 3.0 epochs.
Good For
- Developers requiring a dedicated model for English to Hindi language translation.
- Applications focused on multilingual communication involving these two languages.
- Use cases where a fine-tuned Qwen2.5-3B-Instruct variant for translation is beneficial.