ynklab/Tower-7B-Sep_1c1t

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7BQuant:FP8Context Size:4kTool Calling:SupportedPublished:Sep 7, 2026License:cc-by-nc-sa-4.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Tower-7B-Sep_1c1t is a 7 billion parameter, full-parameter fine-tuned version of Unbabel/TowerInstruct-Mistral-7B-v0.2, developed by ynklab. This model is specifically optimized for multilingual chunk-level machine translation, supporting English to and from German, Spanish, French, Italian, Korean, Dutch, Portuguese, Russian, and Chinese. It utilizes a fixed-range chunking approach (256-512 tokens) and incorporates one preceding source-language context chunk for improved length-consistent document-level translation. The model is ideal for developers requiring contextual machine translation at the document chunk level.

Loading preview...

Overview

Tower-7B-Sep_1c1t is a 7 billion parameter language model developed by ynklab, specifically fine-tuned for multilingual chunk-level machine translation. It is based on the Unbabel/TowerInstruct-Mistral-7B-v0.2 architecture and was trained using the sardinelab/DocBlocks dataset.

Key Capabilities

  • Contextual Machine Translation: Designed for document-level translation, it processes text in fixed-range chunks (256-512 tokens) and utilizes one preceding source-language context chunk to maintain consistency.
  • Multilingual Support: Supports translation between English and nine other languages in both directions: German, Spanish, French, Italian, Korean, Dutch, Portuguese, Russian, and Chinese.
  • Optimized for Doc2FRC: This model is the Sep_1c1t variant from the "Doc2FRC: Length-Consistent Document-Level Machine Translation via Fixed-Range Chunking" paper, with specific inference scripts and prompting available in the Doc2FRC GitHub repository.

Training Details

The model underwent full-parameter supervised fine-tuning for 2 epochs with a maximum sequence length of 32,768 tokens, using bfloat16 precision and an AdamW optimizer.

When to Use This Model

This model is particularly well-suited for applications requiring:

  • Document-level machine translation where context from preceding chunks is crucial.
  • Translating between English and the supported nine languages with a focus on chunk-level consistency.
  • Research and development related to the Doc2FRC methodology for length-consistent translation.

Finetunes
1 models