unconst/Affine-5czsc2fc98-r477-offline-dpo-hialpha-hirank-extrasteps-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-r477-offline-dpo-hialpha-hirank-extrasteps-merged is a 35.1 billion parameter language model developed by unconst, derived from a LoRA-merged checkpoint. This model is a private, internal iteration, specifically noted as a salvage operation from kevin954/Affine-5dfqbbh8ev-sft. With a context length of 32768 tokens, it is intended for internal evaluation and development related to TTL insurance, rather than general public use.

Loading preview...

Model Overview

unconst/Affine-5czsc2fc98-r477-offline-dpo-hialpha-hirank-extrasteps-merged is a 35.1 billion parameter language model with a substantial context length of 32768 tokens. This model represents a LoRA-merged checkpoint, specifically salvaged from kevin954/Affine-5dfqbbh8ev-sft.

Key Characteristics

  • Parameter Count: 35.1 billion parameters, indicating a large-scale model capable of complex language understanding and generation tasks.
  • Context Length: Features a 32768-token context window, allowing it to process and generate very long sequences of text, which is beneficial for tasks requiring extensive contextual awareness.
  • Origin: It is a merged checkpoint, suggesting it incorporates fine-tuning or adaptation layers (LoRA) on top of a base model.
  • Development Status: Described as a "private TTL insurance" model, indicating it is an internal, developmental version, not intended for general release or external use.

Intended Use

This model is explicitly stated as "not a submission until Stage-5 gate clears," highlighting its status as an in-progress, internal asset. Its primary purpose appears to be for internal evaluation and development within the context of "TTL insurance," suggesting specialized applications rather than broad, general-purpose use cases.