Nexesenex commited on
Commit
6eaa5f3
1 Parent(s): 3399aea

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +15 -1
README.md CHANGED
@@ -4,6 +4,18 @@ Q3_K_M to be uploaded shortly.
4
 
5
  Q3_K_S, IQ3_XXS, Q2_K, Q2_K_S, IQ2_XS, IQ2_XXS to follow.
6
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  LlamaCPP Benchs on a non iMatrix Q3_K_M released by Undi95 :
8
 
9
  - Miqu-70B-DPO.q3_k_m.gguf,-,Hellaswag,84.5,400,,2024-02-07 00:00:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,NeverSleep,
@@ -26,4 +38,6 @@ Quite convincing compared to the original Miqu.. with iMatrix :
26
  - Miqu-1-70b-Requant-b1989-iMat-c32_ch400-Q3_K_M.gguf,-,wikitext,4.2957,512,512,2024-01-29 00:00:00,RBF1000000,70b,Mistral_Medium,32768,,,GGUF,- Miqudev,Nexesenex,81
27
  - Miqu-1-70b-Requant-b1989-iMat-c32_ch400-Q3_K_M.gguf,-,wikitext,3.8380,512,512,2024-01-29 00:00:00,RBF1000000,70b,Mistral_Medium,32768,,,GGUF,- Miqudev,Nexesenex,655
28
 
29
- The TQA shows a slight bonus, thanks to the DPO training I believe.
 
 
 
4
 
5
  Q3_K_S, IQ3_XXS, Q2_K, Q2_K_S, IQ2_XS, IQ2_XXS to follow.
6
 
7
+ LlamaCPP Benchs on the Q3_K_M with iMatrix shared here :
8
+
9
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Hellaswag,84.5,,400,2024-02-07 00:00:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
10
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Hellaswag,83.6,,1000,2024-02-07 00:00:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
11
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Arc-Challenge,58.52842809,,299,2024-02-07 05:40:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
12
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Arc-Easy,77.36842105,,570,2024-02-07 05:40:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
13
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,MMLU,49.84025559,,313,2024-02-07 05:40:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
14
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Thruthful-QA,42.83965728,,817,2024-02-07 05:40:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
15
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,Winogrande,78.7687,,1267,2024-02-07 05:40:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,
16
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,wikitext,4.2963,512,512,2024-02-07 00:00:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,81
17
+ - Undi95_Miqu-70B-Alpaca-DPO-b2101-iMat-c32_ch1000-Q3_K_M.gguf,-,wikitext,3.8397,512,512,2024-02-07 00:00:00,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,Nexesenex,655
18
+
19
  LlamaCPP Benchs on a non iMatrix Q3_K_M released by Undi95 :
20
 
21
  - Miqu-70B-DPO.q3_k_m.gguf,-,Hellaswag,84.5,400,,2024-02-07 00:00:00,,70b,Mistral_Medium,32768,,,GGUF,NeverSleep,NeverSleep,
 
38
  - Miqu-1-70b-Requant-b1989-iMat-c32_ch400-Q3_K_M.gguf,-,wikitext,4.2957,512,512,2024-01-29 00:00:00,RBF1000000,70b,Mistral_Medium,32768,,,GGUF,- Miqudev,Nexesenex,81
39
  - Miqu-1-70b-Requant-b1989-iMat-c32_ch400-Q3_K_M.gguf,-,wikitext,3.8380,512,512,2024-01-29 00:00:00,RBF1000000,70b,Mistral_Medium,32768,,,GGUF,- Miqudev,Nexesenex,655
40
 
41
+ The TQA shows a slight bonus, thanks to the DPO training I believe.
42
+ The slightly bonified ARC benchs (a rare thing on DPO releases!) and the respected perplexity show that the model was not dumbified by the DPO training.
43
+ In ST, the models performs beautifully.