unconst/Affine-5czsc2fc98-r489-offline-dpo-hialpha-midrank-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

Affine-5czsc2fc98-r489-offline-dpo-hialpha-midrank-ultraextrasteps-merged is a 35.1 billion parameter language model developed by unconst. This model is a LoRA-merged checkpoint, salvaged from a prior SFT version, and is intended for private TTL insurance purposes. It represents an intermediate stage in its development, not yet cleared for broader submission.

Loading preview...

Model Overview

unconst/Affine-5czsc2fc98-r489-offline-dpo-hialpha-midrank-ultraextrasteps-merged is a substantial 35.1 billion parameter language model with a context length of 32768 tokens. This particular version is a LoRA-merged checkpoint, derived from kevin954/Affine-5dfqbbh8ev-sft.

Key Characteristics

  • Development Stage: This model is explicitly noted as a "salvaged" checkpoint, indicating it's an intermediate or recovery version from an ongoing development process.
  • Purpose: It is designated for "Private TTL insurance," suggesting its current use is internal or for specific, limited validation rather than general deployment.
  • Status: The README states it is "not a submission until Stage-5 gate clears," implying it is still undergoing internal evaluation and has not reached a release-ready state.

When to Consider This Model

Given its current status and stated purpose, this model is primarily relevant for:

  • Internal Development & Research: Users involved in the specific project or research stream that produced this model.
  • Understanding Development Iterations: Researchers interested in observing intermediate checkpoints and LoRA merging techniques within a large language model's lifecycle.

It is important to note that this model is not presented as a general-purpose, production-ready LLM.