ynklab/Qwen2.5-7B-Sep_1c1t

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 7, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

ynklab/Qwen2.5-7B-Sep_1c1t is a 7.6 billion parameter language model, fine-tuned from Qwen/Qwen2.5-7B-Instruct, specifically designed for multilingual chunk-level machine translation. Developed by ynklab, this model excels at translating between English and nine other languages, utilizing a fixed-range chunking approach with one preceding source-language context chunk. It supports a maximum sequence length of 32,768 tokens and is optimized for document-level translation tasks.

Loading preview...

Model Overview

ynklab/Qwen2.5-7B-Sep_1c1t is a 7.6 billion parameter model, fine-tuned from Qwen/Qwen2.5-7B-Instruct, specializing in multilingual chunk-level machine translation. This model is a Sep_1c1t variant, meaning it processes translations by considering one preceding source-language context chunk alongside the current source chunk, translating it into one target chunk. It was developed as part of the Doc2FRC: Length-Consistent Document-Level Machine Translation via Fixed-Range Chunking paper.

Key Capabilities

  • Multilingual Translation: Supports translation between English and German, Spanish, French, Italian, Korean, Dutch, Portuguese, Russian, and Chinese in both directions.
  • Context-Aware Chunking: Utilizes a fixed-range chunking method (256–512 tokens) with a preceding source-language context chunk to maintain document-level consistency.
  • High Context Length: Features a maximum sequence length of 32,768 tokens, suitable for processing longer document segments.
  • Fine-tuned Performance: Achieved through full-parameter supervised fine-tuning on the sardinelab/DocBlocks dataset over 2 epochs.

Good For

  • Document-Level Machine Translation: Ideal for tasks requiring consistent translation across entire documents by leveraging chunk-level context.
  • Research in MT: Useful for researchers exploring fixed-range chunking and context-aware translation methods, particularly those interested in the Doc2FRC paper's methodology.
  • Specific Language Pairs: Excellent for applications needing translation between English and the nine supported languages, especially where context preservation is critical.