Triangle104 commited on
Commit
86bf81f
1 Parent(s): c37f73d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +48 -0
README.md CHANGED
@@ -108,6 +108,54 @@ model-index:
108
  This model was converted to GGUF format from [`flammenai/Mahou-1.5-mistral-nemo-12B`](https://huggingface.co/flammenai/Mahou-1.5-mistral-nemo-12B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
109
  Refer to the [original model card](https://huggingface.co/flammenai/Mahou-1.5-mistral-nemo-12B) for more details on the model.
110
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  ## Use with llama.cpp
112
  Install llama.cpp through brew (works on Mac and Linux)
113
 
 
108
  This model was converted to GGUF format from [`flammenai/Mahou-1.5-mistral-nemo-12B`](https://huggingface.co/flammenai/Mahou-1.5-mistral-nemo-12B) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
109
  Refer to the [original model card](https://huggingface.co/flammenai/Mahou-1.5-mistral-nemo-12B) for more details on the model.
110
 
111
+ ---
112
+ Model details:
113
+ -
114
+ Mahou-1.5-mistral-nemo-12B
115
+
116
+ Mahou is designed to provide short messages in a conversational context. It is capable of casual conversation and character roleplay.
117
+ Chat Format
118
+
119
+ This model has been trained to use ChatML format.
120
+
121
+ <|im_start|>system
122
+ {{system}}<|im_end|>
123
+ <|im_start|>{{char}}
124
+ {{message}}<|im_end|>
125
+ <|im_start|>{{user}}
126
+ {{message}}<|im_end|>
127
+
128
+ Roleplay Format
129
+
130
+ Speech without quotes.
131
+ Actions in *asterisks*
132
+
133
+ *leans against wall cooly* so like, i just casted a super strong spell at magician academy today, not gonna lie, felt badass.
134
+
135
+ SillyTavern Settings
136
+
137
+ Use ChatML for the Context Template.
138
+ Enable Instruct Mode.
139
+ Use the Mahou ChatML Instruct preset.
140
+ Use the Mahou Sampler preset.
141
+
142
+ Method
143
+
144
+ ORPO finetuned with 4x H100 for 3 epochs.
145
+ Open LLM Leaderboard Evaluation Results
146
+
147
+ Detailed results can be found here
148
+ Metric Value
149
+ Avg. 26.28
150
+ IFEval (0-Shot) 67.51
151
+ BBH (3-Shot) 36.26
152
+ MATH Lvl 5 (4-Shot) 5.06
153
+ GPQA (0-shot) 3.47
154
+ MuSR (0-shot) 16.47
155
+ MMLU-PRO (5-shot) 28.91
156
+
157
+ ---
158
+
159
  ## Use with llama.cpp
160
  Install llama.cpp through brew (works on Mac and Linux)
161