Stable-DiffCoder 8B Instruct: instruction-tuned code model from ByteDance
ByteDance Seed published Stable-DiffCoder-8B-Instruct, the instruction-tuned variant of Stable-DiffCoder-8B-Base, on Hugging Face. It is fine-tuned from the base model for conversational code generation tasks using a Llama architecture under the MIT license.
PUBLISHED2026-01-15
OBSERVED2026-08-11
AGE7mo
SOURCES1
- Pipeline: text-generation, conversational
- Library: transformers with custom_code
- Base model: ByteDance-Seed/Stable-DiffCoder-8B-Base (fine-tuned)
- Architecture: llama
- License: MIT
- Deployment: text-generation-inference, endpoints_compatible
- Referenced paper: arxiv:2601.15892
- Repository created 2026-01-15, last modified 2026-03-25
COMMUNITY
No curated reactions recorded for this event. Facts and takes are kept in separate layers — community context is added by hand, never blended into the record above.