Hugging Face's logo

Spaces:
SamsungResearch
/
TRUEBench
Running

App Files Community
4
j_yoon.song commited on
Commit
057c5b1
·
1 Parent(s): 289ccbe

Fix typo

Browse files
Files changed (3) hide show
  1. src/data/length_data.json +1 -1
  2. src/data/stats.csv +1 -1
  3. src/data/stats_lang.csv +1 -1
src/data/length_data.json CHANGED
@@ -951,7 +951,7 @@
951
  "Med Resp": 1318.5
952
  }
953
  },
954
- "Mistral-Samll-3-2 24B-Instruct-2506": {
955
  "Overall": {
956
  "Min": 1,
957
  "Max": 65516,
 
951
  "Med Resp": 1318.5
952
  }
953
  },
954
+ "Mistral-Small-3-2 24B-Instruct-2506": {
955
  "Overall": {
956
  "Min": 1,
957
  "Max": 65516,
src/data/stats.csv CHANGED
@@ -48,7 +48,7 @@ top-p: 1.0" "GPT" "953.5" "326.0" "21.0" "Open" "Think" "On" "41.18" "52.0" "40.
48
  top-p: 0.95" "Gemma" "380.0" "380.0" "27.0" "Open" "Instruct" "Off" "40.86" "44.25" "45.0" "45.82" "36.78" "31.61" "32.23" "22.86" "57.14" "32.87" "39.93"
49
  "Tongyi DeepResearch 30B A3B" "https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B" "temperature: 0.6
50
  top-p: 0.95" "Alibaba" "1147.0" "408.0" "30.0" "Open" "Think" "On" "40.1" "41.25" "33.12" "62.15" "68.18" "44.25" "23.97" "18.57" "41.67" "26.12" "29.19"
51
- "Mistral-Samll-3-2 24B-Instruct-2506" "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506" "temperature: 0.15
52
  top-p: 0.95" "mistralai" "369.0" "369.0" "24.0" "Open" "Instruct" "Off" "39.09" "43.0" "44.69" "43.43" "51.65" "25.86" "22.31" "25.71" "51.98" "31.18" "30.2"
53
  "K2-Think" "https://huggingface.co/LLM360/K2-Think" "temperature: 1.0
54
  top-p: 0.95" "LLM360" "1835.0" "486.0" "32.8" "Open" "Think" "On" "35.06" "35.5" "36.56" "56.18" "47.11" "35.06" "14.05" "12.86" "49.21" "21.63" "23.15"
 
48
  top-p: 0.95" "Gemma" "380.0" "380.0" "27.0" "Open" "Instruct" "Off" "40.86" "44.25" "45.0" "45.82" "36.78" "31.61" "32.23" "22.86" "57.14" "32.87" "39.93"
49
  "Tongyi DeepResearch 30B A3B" "https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B" "temperature: 0.6
50
  top-p: 0.95" "Alibaba" "1147.0" "408.0" "30.0" "Open" "Think" "On" "40.1" "41.25" "33.12" "62.15" "68.18" "44.25" "23.97" "18.57" "41.67" "26.12" "29.19"
51
+ "Mistral-Small-3-2 24B-Instruct-2506" "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506" "temperature: 0.15
52
  top-p: 0.95" "mistralai" "369.0" "369.0" "24.0" "Open" "Instruct" "Off" "39.09" "43.0" "44.69" "43.43" "51.65" "25.86" "22.31" "25.71" "51.98" "31.18" "30.2"
53
  "K2-Think" "https://huggingface.co/LLM360/K2-Think" "temperature: 1.0
54
  top-p: 0.95" "LLM360" "1835.0" "486.0" "32.8" "Open" "Think" "On" "35.06" "35.5" "36.56" "56.18" "47.11" "35.06" "14.05" "12.86" "49.21" "21.63" "23.15"
src/data/stats_lang.csv CHANGED
@@ -48,7 +48,7 @@ top-p: 1.0" "GPT" "953.5" "326.0" "21.0" "Open" "Think" "On" "41.18" "36.67" "42
48
  top-p: 0.95" "Gemma" "380.0" "380.0" "27.0" "Open" "Instruct" "Off" "40.86" "34.44" "35.0" "37.35" "43.9" "42.22" "43.65" "47.49" "41.08" "44.02" "53.55" "39.33" "40.24"
49
  "Tongyi DeepResearch 30B A3B" "https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B" "temperature: 0.6
50
  top-p: 0.95" "Alibaba" "1147.0" "408.0" "30.0" "Open" "Think" "On" "40.1" "36.11" "40.83" "43.37" "44.51" "32.78" "37.02" "44.69" "38.92" "43.48" "46.45" "37.08" "39.63"
51
- "Mistral-Samll-3-2 24B-Instruct-2506" "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506" "temperature: 0.15
52
  top-p: 0.95" "mistralai" "369.0" "369.0" "24.0" "Open" "Instruct" "Off" "39.09" "31.39" "40.0" "36.75" "42.07" "34.44" "44.2" "41.9" "42.16" "45.65" "40.98" "37.64" "38.41"
53
  "K2-Think" "https://huggingface.co/LLM360/K2-Think" "temperature: 1.0
54
  top-p: 0.95" "LLM360" "1835.0" "486.0" "32.8" "Open" "Think" "On" "35.06" "29.17" "36.11" "30.12" "44.51" "26.67" "33.15" "38.55" "37.84" "41.85" "37.7" "33.71" "36.59"
 
48
  top-p: 0.95" "Gemma" "380.0" "380.0" "27.0" "Open" "Instruct" "Off" "40.86" "34.44" "35.0" "37.35" "43.9" "42.22" "43.65" "47.49" "41.08" "44.02" "53.55" "39.33" "40.24"
49
  "Tongyi DeepResearch 30B A3B" "https://huggingface.co/Alibaba-NLP/Tongyi-DeepResearch-30B-A3B" "temperature: 0.6
50
  top-p: 0.95" "Alibaba" "1147.0" "408.0" "30.0" "Open" "Think" "On" "40.1" "36.11" "40.83" "43.37" "44.51" "32.78" "37.02" "44.69" "38.92" "43.48" "46.45" "37.08" "39.63"
51
+ "Mistral-Small-3-2 24B-Instruct-2506" "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506" "temperature: 0.15
52
  top-p: 0.95" "mistralai" "369.0" "369.0" "24.0" "Open" "Instruct" "Off" "39.09" "31.39" "40.0" "36.75" "42.07" "34.44" "44.2" "41.9" "42.16" "45.65" "40.98" "37.64" "38.41"
53
  "K2-Think" "https://huggingface.co/LLM360/K2-Think" "temperature: 1.0
54
  top-p: 0.95" "LLM360" "1835.0" "486.0" "32.8" "Open" "Think" "On" "35.06" "29.17" "36.11" "30.12" "44.51" "26.67" "33.15" "38.55" "37.84" "41.85" "37.7" "33.71" "36.59"