update readme

README.md

@@ -46,11 +46,19 @@ This repository provides a Japanese GPT-NeoX model of 3.6 billion parameters. Th
     The RL data is the subset of the following dataset and has been translated into Japanese.
     * [Anthropic HH RLHF data](https://huggingface.co/datasets/Anthropic/hh-rlhf)
 
+* **Model Series**
+
+    | Variant | Link |
+    | :-- | :--|
+    | 3.6B PPO | https://huggingface.co/rinna/japanese-gpt-neox-3.6b-instruction-ppo |
+    | 3.6B SFT-v2 | https://huggingface.co/rinna/japanese-gpt-neox-3.6b-instruction-sft-v2 |
+    | 3.6B SFT | https://huggingface.co/rinna/japanese-gpt-neox-3.6b-instruction-sft |
+    | 3.6B pretrained | https://huggingface.co/rinna/japanese-gpt-neox-3.6b |
+    
 * **Authors**
     
     [Tianyu Zhao](https://huggingface.co/tianyuz) and [Kei Sawada](https://huggingface.co/keisawada)
 
-
 # Limitations
 * We found this verison of PPO model tends to generate repeated text more often than its SFT counterpart, and thus we set `repetition_penalty=1.1` for better generation performance. (*The same generation hyper-parameters are applied to the SFT model in aforementioned evaluation experiments.*) You can also explore other hyperparameter combinations that yield higher generation randomness/diversity for better generation quality, e.g. `temperature=0.9, repetition_penalty=1.0`.