unconst/Affine-5czsc2fc98-r205-merged

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 14, 2026Architecture:Transformer Featherless Exclusive Cold

The unconst/Affine-5czsc2fc98-r205-merged model is a 35.1 billion parameter language model, derived from a LoRA merge of kevin954/Affine-5dfqbbh8ev-sft. This model is currently a private checkpoint under TTL insurance, indicating it is in an experimental or pre-release stage. Its primary characteristic is being a merged checkpoint, suggesting it combines different fine-tuning or adaptation layers. Developers should note its experimental status and origin from a LoRA merge.

Loading preview...

Model Overview

The unconst/Affine-5czsc2fc98-r205-merged is a substantial 35.1 billion parameter language model. It is specifically noted as a "merged checkpoint salvage," indicating its origin from a LoRA (Low-Rank Adaptation) merge of the kevin954/Affine-5dfqbbh8ev-sft model.

Key Characteristics

  • Parameter Count: 35.1 billion parameters, suggesting a high capacity for complex language understanding and generation tasks.
  • Context Length: Supports a context window of 32768 tokens, enabling processing of extensive inputs and generating coherent long-form content.
  • Development Status: Described as a "private TTL insurance" checkpoint, meaning it is in an experimental or pre-release phase and not yet a final submission. This implies ongoing development and potential for changes.
  • Origin: The model is a result of a LoRA merge, a technique often used to efficiently adapt large pre-trained models to new tasks or datasets without full fine-tuning.

Potential Use Cases

Given its experimental status and origin, this model is currently best suited for:

  • Research and Development: Exploring the effects of LoRA merges on large language models.
  • Internal Testing: Evaluating its performance on specific tasks before a public release.
  • Advanced Prototyping: For developers comfortable with potentially unstable or evolving models.