unconst/Affine-5czsc2fc98-r520-offline-dpo-hialpha-midrank-lobeta-ultraextrasteps-merged

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 16, 2026Architecture:Transformer Featherless Exclusive Cold

The unconst/Affine-5czsc2fc98-r520-offline-dpo-hialpha-midrank-lobeta-ultraextrasteps-merged model is a 35.1 billion parameter language model, merged from kevin954/Affine-5dfqbbh8ev-sft. It features a 32768 token context length and is a LoRA-merged checkpoint. This model is intended for internal evaluation and development, serving as a salvage checkpoint rather than a final submission.

Loading preview...

Model Overview

The unconst/Affine-5czsc2fc98-r520-offline-dpo-hialpha-midrank-lobeta-ultraextrasteps-merged is a substantial language model with 35.1 billion parameters and a context length of 32768 tokens. This model is a LoRA-merged checkpoint, derived from the kevin954/Affine-5dfqbbh8ev-sft base model.

Key Characteristics

  • Parameter Count: 35.1 billion, indicating a large capacity for complex language understanding and generation tasks.
  • Context Length: Supports an extensive context window of 32768 tokens, allowing it to process and generate longer, more coherent texts.
  • Origin: It is a LoRA-merged checkpoint, suggesting it has undergone fine-tuning or adaptation from a previous version.

Intended Use

This model is described as a "merged checkpoint salvage" and is explicitly noted as "not a submission until Stage-5 gate clears." This indicates its primary purpose is for internal development, evaluation, and potentially as an intermediate step in a larger model development pipeline. It is not presented as a ready-for-production model but rather as a developmental asset for further refinement or analysis.