AI BriefingIFM Releases K2-Horizon-7B
Key point
IFM has released K2-Horizon-7B, which combines a diffusion adapter with a causal LLM, claiming inference speeds of up to 5200 tps and lossless performance.
Details
IFM has released K2-Horizon-7B, which combines a plug-and-play diffusion adapter with a causal LLM architecture. This model applies a hybrid approach that uses a diffusion adapter alongside existing autoregressive weights to increase inference speed.
Key performance metrics are as follows:
- Inference Speed: Achieves up to 5200 tps (tokens per second)
- Quality Maintenance: Claims lossless performance despite the speed improvements
The model adopts an approach that utilizes diffusion models as a supplementary component to overcome the structural limitations of existing LLMs, and the related research has been published on arXiv.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Inception Releases Ultra-Fast LLM 'Mercury 2.5' Generating 770 Tokens Per Second