joshycodes/qwen3-32b-control-B1-sdf
TEXT GENERATIONPricing:Input $0.408 / Cached $0.0816 / Output $1.972Concurrent Unit Cost:2Model Size:32BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 22, 2026License:otherArchitecture:Transformer Featherless Exclusive Cold
joshycodes/qwen3-32b-control-B1-sdf is a 32 billion parameter Qwen3-based language model that underwent continued pretraining on a self-authored corpus. This model was trained for one epoch on 32.4 million tokens, focusing on how its character was formed and how SDF works. It represents a research checkpoint for developing subsequent model versions, with its capabilities, alignment, and identity yet to be evaluated.
Loading preview...
Model Overview
joshycodes/qwen3-32b-control-B1-sdf is a 32 billion parameter model derived from the Qwen3-32B architecture. This version is a research checkpoint that has undergone a unique continued pretraining phase.
Key Characteristics
- Self-Authored Corpus Pretraining: The model was continued-pretrained on a corpus it generated itself, specifically focusing on its own character development and the principles of SDF (Signed Distance Function).
- Training Details: The pretraining involved one epoch over 32,448,549 tokens from 40,842 documents, all of which were ordinary text, with no self-authored content in the initial corpus.
- Developmental Stage: This model is explicitly noted as a research checkpoint. It has not yet been evaluated for its general capabilities, alignment, or identity.
Intended Use and Limitations
- Research Focus: This model is primarily intended for research purposes, specifically to explore the effects of continued pretraining on a self-authored corpus for future model iterations.
- Undeployed Status: Users are advised not to deploy this model for production or critical applications, as its performance, safety, and alignment characteristics are currently unassessed.