wyy-code commited on
Commit
877e0f3
·
verified ·
1 Parent(s): 037a276

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -27,7 +27,7 @@ Welcome to **InfiX-ai**! We are a research-driven organization dedicated to buil
27
  #### Model over Models
28
  - [InfiFusion](https://huggingface.co/collections/InfiX-ai/infifusion-683c7d7f00c71614ba8ceb96): **InfiFusion** is a logit-level fusion pipeline based on Universal Logit Distillation, enhanced with Top-K filtering and logits standardization. It supports both pairwise and unified fusion strategies to balance performance and efficiency.
29
  - [InfiGFusion](https://huggingface.co/InfiX-ai/InfiGFusion-14B): **InfiGFusion** is a structure-aware extension that builds co-activation graphs from logits and aligns them via an efficient Gromov-Wasserstein loss approximation, capturing cross-dimension semantic dependencies for stronger reasoning.
30
- - [InfiFPO](https://huggingface.co/InfiX-ai/InfiFPO-14B): **InfiFPO** is a lightweight fusion method for preference alignment that injects fused model behavior into reward optimization, enabling richer signal during DPO-style fine-tuning.
31
 
32
  ---
33
  ### News📰
 
27
  #### Model over Models
28
  - [InfiFusion](https://huggingface.co/collections/InfiX-ai/infifusion-683c7d7f00c71614ba8ceb96): **InfiFusion** is a logit-level fusion pipeline based on Universal Logit Distillation, enhanced with Top-K filtering and logits standardization. It supports both pairwise and unified fusion strategies to balance performance and efficiency.
29
  - [InfiGFusion](https://huggingface.co/InfiX-ai/InfiGFusion-14B): **InfiGFusion** is a structure-aware extension that builds co-activation graphs from logits and aligns them via an efficient Gromov-Wasserstein loss approximation, capturing cross-dimension semantic dependencies for stronger reasoning.
30
+ - [InfiFPO](https://huggingface.co/InfiX-ai/InfiFPO-14B): **InfiFPO** is a lightweight fusion method during the preference alignment phase that injects fused model behavior into preference learning, enabling richer signal during DPO-style fine-tuning.
31
 
32
  ---
33
  ### News📰