JosephusCheung commited on
Commit
5e04d1c
·
1 Parent(s): 2a97280

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +9 -6
README.md CHANGED
@@ -38,7 +38,7 @@ tags:
38
  # CausalLM 14B - Fully Compatible with Meta LLaMA 2
39
  Use the transformers library that does not require remote/external code to load the model, AutoModelForCausalLM and AutoTokenizer (or manually specify LlamaForCausalLM to load LM, GPT2Tokenizer to load Tokenizer), and model quantization is fully compatible with GGUF (llama.cpp), GPTQ, and AWQ.
40
 
41
- **News: SOTA chat model of its size on 🤗 Open LLM Leaderboard**
42
 
43
  # Recent Updates: [DPO-α Version](https://huggingface.co/CausalLM/14B-DPO-alpha) outperforms Zephyr-β in MT-Bench
44
 
@@ -133,12 +133,15 @@ We are currently unable to produce accurate benchmark templates for non-QA tasks
133
  ## 🤗 Open LLM Leaderboard
134
  SOTA chat model of its size on 🤗 Open LLM Leaderboard.
135
 
136
- ![leaderboard](https://huggingface.co/JosephusCheung/tmp/resolve/main/leaderboard.png)
 
 
 
137
 
138
  # 因果语言模型 14B - 与 Meta LLaMA 2 完全兼容
139
  使用无需远程/外部代码的transformers库加载模型,AutoModelForCausalLM和AutoTokenizer(或者手动指定LlamaForCausalLM加载LM, GPT2Tokenizer加载Tokenizer),并且模型量化与GGUF(llama.cpp)、GPTQ、AWQ完全兼容。
140
 
141
- # 新消息:🤗 Open LLM 排行榜上同尺寸的聊天模型中评分最高
142
 
143
  # 最近更新: [DPO-α Version](https://huggingface.co/CausalLM/14B-DPO-alpha) 在 MT-Bench 超过 Zephyr-β
144
 
@@ -231,6 +234,6 @@ STEM准确率:66.71
231
  *JCommonsenseQA 基准测试结果非常非常接近 [Japanese Stable LM Gamma 7B (83.47)](https://github.com/Stability-AI/lm-evaluation-harness/tree/jp-stable),当前 SOTA 日文 LM 。然而,我们的模型并未在日文上进行特别的大量文本训练。这似乎能体现元语言的跨语言迁移能力。*
232
 
233
  ## 🤗 Open LLM 排行榜
234
- 🤗 Open LLM 排行榜上同尺寸的聊天模型中评分最高
235
-
236
- ![leaderboard](https://huggingface.co/JosephusCheung/tmp/resolve/main/leaderboard.png)
 
38
  # CausalLM 14B - Fully Compatible with Meta LLaMA 2
39
  Use the transformers library that does not require remote/external code to load the model, AutoModelForCausalLM and AutoTokenizer (or manually specify LlamaForCausalLM to load LM, GPT2Tokenizer to load Tokenizer), and model quantization is fully compatible with GGUF (llama.cpp), GPTQ, and AWQ.
40
 
41
+ **News: DPO ver. Rank #1 ~13B - SOTA model of its size on 🤗 Open LLM Leaderboard**
42
 
43
  # Recent Updates: [DPO-α Version](https://huggingface.co/CausalLM/14B-DPO-alpha) outperforms Zephyr-β in MT-Bench
44
 
 
133
  ## 🤗 Open LLM Leaderboard
134
  SOTA chat model of its size on 🤗 Open LLM Leaderboard.
135
 
136
+ Dec 3, 2023
137
+ DPO Version Rank **#1** non-base model, of its size on 🤗 Open LLM Leaderboard, outperforms **ALL** ~13B chat models.
138
+
139
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/63468a143ea42ee2cb49ddd1/8nV0yOTteP208bjbCv5MC.png)
140
 
141
  # 因果语言模型 14B - 与 Meta LLaMA 2 完全兼容
142
  使用无需远程/外部代码的transformers库加载模型,AutoModelForCausalLM和AutoTokenizer(或者手动指定LlamaForCausalLM加载LM, GPT2Tokenizer加载Tokenizer),并且模型量化与GGUF(llama.cpp)、GPTQ、AWQ完全兼容。
143
 
144
+ # 新消息:DPO 版本在~13B排名第1 🤗 Open LLM 排行榜上同尺寸的所有模型中评分最高
145
 
146
  # 最近更新: [DPO-α Version](https://huggingface.co/CausalLM/14B-DPO-alpha) 在 MT-Bench 超过 Zephyr-β
147
 
 
234
  *JCommonsenseQA 基准测试结果非常非常接近 [Japanese Stable LM Gamma 7B (83.47)](https://github.com/Stability-AI/lm-evaluation-harness/tree/jp-stable),当前 SOTA 日文 LM 。然而,我们的模型并未在日文上进行特别的大量文本训练。这似乎能体现元语言的跨语言迁移能力。*
235
 
236
  ## 🤗 Open LLM 排行榜
237
+ Dec 3, 2023
238
+ DPO版本在🤗 Open LLM 排行榜上~13B的**所有**聊天模型中**排名第1**
239
+ ![image/png](https://cdn-uploads.huggingface.co/production/uploads/63468a143ea42ee2cb49ddd1/8nV0yOTteP208bjbCv5MC.png)