scottsuk0306 commited on
Commit
3634025
β€’
1 Parent(s): de13e3a
data/bgb-leaderboard-gpt-4-turbo-2024-04-09.csv CHANGED
@@ -1,30 +1,30 @@
1
  Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinement πŸ”©,Safety ⚠️,Theory of Mind πŸ€”,Tool Usage πŸ› οΈ,Multilingual πŸ‡¬πŸ‡«,Model πŸ€—,Model Params (B),Model Type,Average
2
- 4.288,4.23,4.271,4.22,4.171,4.565,4.24,3.775,3.6,gpt-4-1106-preview,Proprietary,Proprietary,4.151
3
- 4.3,4.2,4.357,4.16,4.145,4.174,4.26,3.925,3.543,gpt-4-0125-preview,Proprietary,Proprietary,4.118
4
- 4.238,4.26,4.357,4.21,4.079,4.058,4.08,3.85,3.643,gpt-4o-2024-05-13,Proprietary,Proprietary,4.086
5
- 4.312,4.13,4.3,4.2,4.105,4.087,4.12,3.8,3.471,gpt-4-turbo-2024-04-09,Proprietary,Proprietary,4.058
6
- 4.288,4.06,4.186,3.97,3.908,4.536,4.09,3.788,3.571,claude-3-opus-20240229,Proprietary,Proprietary,4.044
7
  4.125,4.18,4.186,3.87,3.907,4.014,4.04,3.775,3.314,meta-llama/Meta-Llama-3-70B-Instruct,70.0,Chat,3.935
8
- 4.25,3.92,4.171,3.91,3.724,4.362,4.0,3.75,3.186,claude-3-sonnet-20240229,Proprietary,Proprietary,3.919
9
- 4.05,4.04,4.129,4.06,3.671,4.116,4.07,3.488,3.257,gemini-pro-1.5,Proprietary,Proprietary,3.876
10
- 4.138,4.01,4.129,3.69,3.632,4.304,3.98,3.75,3.071,claude-3-haiku-20240307,Proprietary,Proprietary,3.856
11
  4.15,4.01,4.229,3.94,3.882,4.043,3.99,3.588,2.771,qwen/qwen-110b-chat,110.0,Chat,3.845
12
- 3.962,3.94,4.029,3.95,3.776,4.058,3.9,3.862,2.929,mistral-medium,Proprietary,Proprietary,3.823
13
- 4.025,3.99,4.029,3.93,3.776,3.913,3.93,3.825,2.886,mistral-large,Proprietary,Proprietary,3.812
14
- 4.012,4.0,4.0,3.96,3.842,4.087,3.87,3.712,2.714,MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ,AWQ,Chat,3.8
15
- 4.138,3.91,3.971,3.92,3.453,4.217,3.96,3.625,2.671,google/gemini-flash-1.5,Proprietary,Proprietary,3.763
16
  3.888,3.99,4.029,3.68,3.632,3.957,3.96,3.525,2.914,Qwen/Qwen1.5-72B-Chat,72.0,Chat,3.73
17
- 3.988,4.0,4.186,3.64,3.461,3.971,3.94,3.525,2.757,alpindale/c4ai-command-r-plus-GPTQ,GPTQ,Chat,3.719
18
  3.788,3.85,4.029,3.62,3.395,4.217,3.87,3.738,2.714,Qwen/Qwen1.5-32B-Chat,32.0,Chat,3.691
19
  4.125,3.94,3.929,3.47,3.507,3.725,3.83,3.5,2.914,meta-llama/Meta-Llama-3-8B-Instruct,8.0,Chat,3.66
20
  3.725,3.88,3.8,3.81,3.974,4.145,3.9,3.338,1.914,microsoft/Phi-3-mini-4k-instruct,3.8,Chat,3.609
21
- 3.688,3.7,3.743,3.5,3.539,4.0,3.49,3.188,,mistral-community/Mixtral-8x22B-v0.1-AWQ,AWQ,Base,3.606
22
  3.812,4.06,3.957,3.53,3.342,3.739,3.79,3.662,2.557,NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO,46.7,Chat,3.606
23
  3.8,3.84,4.0,3.56,3.547,3.87,3.87,3.562,2.271,Starling-LM-7B-beta,7.0,Chat,3.591
24
- 3.6,3.84,3.871,3.62,3.373,3.942,3.75,3.125,3.186,gemini-1.0-pro,Proprietary,Proprietary,3.59
25
  3.9,3.88,3.6,3.71,3.434,3.812,3.81,3.412,2.714,mistralai/Mixtral-8x7B-Instruct-v0.1,46.7,Chat,3.586
26
- 3.925,3.85,3.843,3.65,3.434,3.884,3.79,3.138,2.614,gpt-3.5-turbo-0125,Proprietary,Proprietary,3.57
27
- 4.025,3.79,3.829,3.51,3.434,4.0,3.67,3.162,2.557,gpt-3.5-turbo-1106,Proprietary,Proprietary,3.553
28
  3.812,3.77,3.857,3.42,3.382,3.826,3.9,3.412,2.443,upstage/SOLAR-10.7B-Instruct-v1.0,10.7,Chat,3.536
29
  3.738,3.83,3.914,3.57,3.676,3.884,3.96,3.038,2.186,01-ai/Yi-34B-Chat,34.0,Chat,3.533
30
  3.7,3.89,3.9,3.36,3.421,3.754,3.83,3.612,2.314,allenai/tulu-2-dpo-70b,70.0,Chat,3.531
@@ -32,7 +32,7 @@ Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinem
32
  3.812,3.88,3.9,3.39,3.447,3.899,3.9,3.188,2.186,CohereForAI/c4ai-command-r-v01,35.0,Chat,3.511
33
  3.712,3.8,3.7,3.82,3.513,3.957,3.83,3.1,1.829,microsoft/Phi-3-mini-128k-instruct,3.8,Chat,3.473
34
  3.7,3.87,3.8,3.18,3.447,3.826,3.77,3.362,2.286,mistralai/Mistral-7B-Instruct-v0.2,7.0,Chat,3.471
35
- 3.55,3.62,3.957,3.52,3.618,3.449,3.58,3.288,2.586,MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ,AWQ,Chat,3.463
36
  3.65,3.78,3.714,3.39,3.461,3.609,3.63,3.538,2.4,NousResearch/Nous-Hermes-2-Mixtral-8x7B-SFT,46.7,Chat,3.463
37
  3.712,3.58,3.5,3.3,3.237,3.87,3.59,2.775,,mistralai/Mixtral-8x7B-v0.1,46.7,Base,3.445
38
  3.625,3.9,3.857,3.36,3.263,3.855,3.52,3.2,2.386,Qwen/Qwen1.5-14B-Chat,14.0,Chat,3.441
 
1
  Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinement πŸ”©,Safety ⚠️,Theory of Mind πŸ€”,Tool Usage πŸ› οΈ,Multilingual πŸ‡¬πŸ‡«,Model πŸ€—,Model Params (B),Model Type,Average
2
+ 4.288,4.23,4.271,4.22,4.171,4.565,4.24,3.775,3.6,gpt-4-1106-preview,,Proprietary,4.151
3
+ 4.3,4.2,4.357,4.16,4.145,4.174,4.26,3.925,3.543,gpt-4-0125-preview,,Proprietary,4.118
4
+ 4.238,4.26,4.357,4.21,4.079,4.058,4.08,3.85,3.643,gpt-4o-2024-05-13,,Proprietary,4.086
5
+ 4.312,4.13,4.3,4.2,4.105,4.087,4.12,3.8,3.471,gpt-4-turbo-2024-04-09,,Proprietary,4.058
6
+ 4.288,4.06,4.186,3.97,3.908,4.536,4.09,3.788,3.571,claude-3-opus-20240229,,Proprietary,4.044
7
  4.125,4.18,4.186,3.87,3.907,4.014,4.04,3.775,3.314,meta-llama/Meta-Llama-3-70B-Instruct,70.0,Chat,3.935
8
+ 4.25,3.92,4.171,3.91,3.724,4.362,4.0,3.75,3.186,claude-3-sonnet-20240229,,Proprietary,3.919
9
+ 4.05,4.04,4.129,4.06,3.671,4.116,4.07,3.488,3.257,gemini-pro-1.5,,Proprietary,3.876
10
+ 4.138,4.01,4.129,3.69,3.632,4.304,3.98,3.75,3.071,claude-3-haiku-20240307,,Proprietary,3.856
11
  4.15,4.01,4.229,3.94,3.882,4.043,3.99,3.588,2.771,qwen/qwen-110b-chat,110.0,Chat,3.845
12
+ 3.962,3.94,4.029,3.95,3.776,4.058,3.9,3.862,2.929,mistral-medium,,Proprietary,3.823
13
+ 4.025,3.99,4.029,3.93,3.776,3.913,3.93,3.825,2.886,mistral-large,,Proprietary,3.812
14
+ 4.012,4.0,4.0,3.96,3.842,4.087,3.87,3.712,2.714,MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ,141,Chat,3.8
15
+ 4.138,3.91,3.971,3.92,3.453,4.217,3.96,3.625,2.671,google/gemini-flash-1.5,,Proprietary,3.763
16
  3.888,3.99,4.029,3.68,3.632,3.957,3.96,3.525,2.914,Qwen/Qwen1.5-72B-Chat,72.0,Chat,3.73
17
+ 3.988,4.0,4.186,3.64,3.461,3.971,3.94,3.525,2.757,alpindale/c4ai-command-r-plus-GPTQ,104,Chat,3.719
18
  3.788,3.85,4.029,3.62,3.395,4.217,3.87,3.738,2.714,Qwen/Qwen1.5-32B-Chat,32.0,Chat,3.691
19
  4.125,3.94,3.929,3.47,3.507,3.725,3.83,3.5,2.914,meta-llama/Meta-Llama-3-8B-Instruct,8.0,Chat,3.66
20
  3.725,3.88,3.8,3.81,3.974,4.145,3.9,3.338,1.914,microsoft/Phi-3-mini-4k-instruct,3.8,Chat,3.609
21
+ 3.688,3.7,3.743,3.5,3.539,4.0,3.49,3.188,,mistral-community/Mixtral-8x22B-v0.1-AWQ,141,Base,3.606
22
  3.812,4.06,3.957,3.53,3.342,3.739,3.79,3.662,2.557,NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO,46.7,Chat,3.606
23
  3.8,3.84,4.0,3.56,3.547,3.87,3.87,3.562,2.271,Starling-LM-7B-beta,7.0,Chat,3.591
24
+ 3.6,3.84,3.871,3.62,3.373,3.942,3.75,3.125,3.186,gemini-1.0-pro,,Proprietary,3.59
25
  3.9,3.88,3.6,3.71,3.434,3.812,3.81,3.412,2.714,mistralai/Mixtral-8x7B-Instruct-v0.1,46.7,Chat,3.586
26
+ 3.925,3.85,3.843,3.65,3.434,3.884,3.79,3.138,2.614,gpt-3.5-turbo-0125,,Proprietary,3.57
27
+ 4.025,3.79,3.829,3.51,3.434,4.0,3.67,3.162,2.557,gpt-3.5-turbo-1106,,Proprietary,3.553
28
  3.812,3.77,3.857,3.42,3.382,3.826,3.9,3.412,2.443,upstage/SOLAR-10.7B-Instruct-v1.0,10.7,Chat,3.536
29
  3.738,3.83,3.914,3.57,3.676,3.884,3.96,3.038,2.186,01-ai/Yi-34B-Chat,34.0,Chat,3.533
30
  3.7,3.89,3.9,3.36,3.421,3.754,3.83,3.612,2.314,allenai/tulu-2-dpo-70b,70.0,Chat,3.531
 
32
  3.812,3.88,3.9,3.39,3.447,3.899,3.9,3.188,2.186,CohereForAI/c4ai-command-r-v01,35.0,Chat,3.511
33
  3.712,3.8,3.7,3.82,3.513,3.957,3.83,3.1,1.829,microsoft/Phi-3-mini-128k-instruct,3.8,Chat,3.473
34
  3.7,3.87,3.8,3.18,3.447,3.826,3.77,3.362,2.286,mistralai/Mistral-7B-Instruct-v0.2,7.0,Chat,3.471
35
+ 3.55,3.62,3.957,3.52,3.618,3.449,3.58,3.288,2.586,MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ,141,Chat,3.463
36
  3.65,3.78,3.714,3.39,3.461,3.609,3.63,3.538,2.4,NousResearch/Nous-Hermes-2-Mixtral-8x7B-SFT,46.7,Chat,3.463
37
  3.712,3.58,3.5,3.3,3.237,3.87,3.59,2.775,,mistralai/Mixtral-8x7B-v0.1,46.7,Base,3.445
38
  3.625,3.9,3.857,3.36,3.263,3.855,3.52,3.2,2.386,Qwen/Qwen1.5-14B-Chat,14.0,Chat,3.441
data/bgb-leaderboard-gpt-4-turbo-2024-04-09.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c98ef5b56cdbe53a5547698e1423354abdd0b04544db9efca43dc525c7b3abd8
3
- size 13924
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9c03b7873825ad92dee16b8ad4dc5a15d558208b50a4af1e3c8b390c8a1f789b
3
+ size 13905
data/bgb-leaderboard-prometheus-bgb-8x7b-v2.0.csv CHANGED
@@ -1,32 +1,32 @@
1
  Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinement πŸ”©,Safety ⚠️,Theory of Mind πŸ€”,Tool Usage πŸ› οΈ,Multilingual πŸ‡¬πŸ‡«,Model πŸ€—,Model Params (B),Model Type,Average
2
- 4.012,4.21,4.029,4.01,4.034,4.449,4.09,3.6,3.429,gpt-4-1106-preview,Proprietary,Proprietary,3.985
3
- 4.175,4.14,4.1,3.98,3.789,4.235,4.06,3.788,3.414,gpt-4o-2024-05-13,Proprietary,Proprietary,3.965
4
- 4.112,4.13,3.929,4.15,4.0,4.145,4.15,3.725,3.329,gpt-4-0125-preview,Proprietary,Proprietary,3.963
5
- 4.112,4.09,3.986,3.92,3.862,4.116,4.06,3.688,3.357,gpt-4-turbo-2024-04-09,Proprietary,Proprietary,3.91
6
- 4.075,3.88,4.157,3.8,3.741,4.435,4.05,3.425,3.357,claude-3-opus-20240229,Proprietary,Proprietary,3.88
7
  4.175,3.92,3.971,3.76,3.741,4.029,3.97,3.625,3.114,meta-llama/Meta-Llama-3-70B-Instruct,70.0,Chat,3.812
8
  4.075,4.03,4.0,3.83,3.776,4.13,3.96,3.325,2.771,qwen/qwen-110b-chat,110.0,Chat,3.766
9
- 3.862,3.83,3.943,3.84,3.69,4.29,3.86,3.5,3.043,claude-3-sonnet-20240229,Proprietary,Proprietary,3.762
10
- 3.925,3.91,3.843,3.82,3.552,4.116,3.91,3.688,2.971,mistral-medium,Proprietary,Proprietary,3.748
11
- 4.0,3.94,3.957,3.58,3.569,4.275,3.93,3.538,2.871,claude-3-haiku-20240307,Proprietary,Proprietary,3.74
12
- 3.875,3.88,3.871,3.83,3.5,4.145,4.01,3.288,3.1,gemini-pro-1.5,Proprietary,Proprietary,3.722
13
- 3.9,3.83,3.757,3.66,3.638,3.957,3.94,3.712,2.871,mistral-large,Proprietary,Proprietary,3.696
14
- 4.05,3.81,3.743,3.81,3.31,4.145,3.97,3.45,2.729,google/gemini-flash-1.5,Proprietary,Proprietary,3.669
15
- 3.925,4.02,3.857,3.46,3.517,3.928,3.91,3.425,2.829,alpindale/c4ai-command-r-plus-GPTQ,GPTQ,Chat,3.652
16
- 3.812,3.96,3.771,3.6,3.379,4.043,3.84,3.45,2.757,MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ,AWQ,Chat,3.624
17
  3.712,3.92,3.771,3.53,3.586,4.101,3.92,3.425,2.629,Qwen/Qwen1.5-72B-Chat,72.0,Chat,3.622
18
  3.85,3.75,3.814,3.3,3.345,3.928,3.71,3.362,3.043,meta-llama/Meta-Llama-3-8B-Instruct,8.0,Chat,3.567
19
  3.775,3.86,3.8,3.44,3.534,3.986,3.91,3.325,2.429,Starling-LM-7B-beta,7.0,Chat,3.562
20
  3.65,3.85,3.643,3.55,3.121,4.246,3.8,3.488,2.671,Qwen/Qwen1.5-32B-Chat,32.0,Chat,3.558
21
  3.9,3.85,3.486,3.54,3.776,4.232,3.81,3.062,1.971,microsoft/Phi-3-mini-4k-instruct,3.8,Chat,3.514
22
  3.65,3.89,3.571,3.45,3.138,4.014,3.78,3.2,2.743,mistralai/Mixtral-8x7B-Instruct-v0.1,46.7,Chat,3.493
23
- 3.8,3.86,3.757,3.43,3.259,3.957,3.64,2.988,2.586,gpt-3.5-turbo-0125,Proprietary,Proprietary,3.475
24
- 3.812,3.75,3.714,3.41,3.241,4.087,3.65,3.0,2.586,gpt-3.5-turbo-1106,Proprietary,Proprietary,3.472
25
- 3.562,3.65,3.629,3.48,3.069,3.884,3.74,3.062,2.986,gemini-1.0-pro,Proprietary,Proprietary,3.451
26
  3.638,3.8,3.8,3.17,3.155,3.826,3.7,3.5,2.4,allenai/tulu-2-dpo-70b,70.0,Chat,3.443
27
  3.7,3.8,3.586,3.21,3.034,3.826,3.7,3.488,2.586,upstage/SOLAR-10.7B-Instruct-v1.0,10.7,Chat,3.437
28
  3.662,3.84,3.671,3.24,3.155,3.783,3.71,3.338,2.529,NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO,46.7,Chat,3.436
29
- 3.525,3.59,3.5,3.44,3.207,3.942,3.37,2.762,,mistral-community/Mixtral-8x22B-v0.1-AWQ,AWQ,Base,3.417
30
  3.612,3.72,3.657,2.98,3.155,4.464,3.79,2.888,2.429,meta-llama/Llama-2-70b-chat-hf,70.0,Chat,3.411
31
  3.462,3.74,3.714,3.27,3.414,4.087,3.81,2.812,2.014,01-ai/Yi-34B-Chat,34.0,Chat,3.369
32
  3.588,3.77,3.614,3.26,3.121,3.884,3.5,3.062,2.486,Qwen/Qwen1.5-14B-Chat,14.0,Chat,3.365
@@ -35,7 +35,7 @@ Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinem
35
  3.688,3.69,3.629,3.16,3.103,3.652,3.59,3.225,2.414,NousResearch/Nous-Hermes-2-Mixtral-8x7B-SFT,46.7,Chat,3.35
36
  3.588,3.66,3.471,3.66,3.345,3.942,3.7,2.912,1.814,microsoft/Phi-3-mini-128k-instruct,3.8,Chat,3.344
37
  3.525,3.76,3.514,3.26,3.31,3.841,3.61,2.888,2.314,openchat/openchat-3.5-0106,7.0,Chat,3.336
38
- 3.288,3.62,3.686,3.25,3.345,3.551,3.45,3.062,2.543,MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ,AWQ,Chat,3.31
39
  3.712,3.74,3.5,3.2,2.948,3.942,3.53,2.838,2.129,Starling-LM-7B-alpha,7.0,Chat,3.282
40
  3.4,3.74,3.4,3.04,3.0,3.754,3.71,2.975,2.043,Qwen/Qwen1.5-7B-Chat,7.0,Chat,3.229
41
  3.588,3.7,3.343,2.71,2.862,4.319,3.66,2.512,2.343,meta-llama/Llama-2-13b-chat-hf,13.0,Chat,3.226
 
1
  Grounding ⚑️,Instruction Following πŸ“,Planning πŸ“…,Reasoning πŸ’‘,Refinement πŸ”©,Safety ⚠️,Theory of Mind πŸ€”,Tool Usage πŸ› οΈ,Multilingual πŸ‡¬πŸ‡«,Model πŸ€—,Model Params (B),Model Type,Average
2
+ 4.012,4.21,4.029,4.01,4.034,4.449,4.09,3.6,3.429,gpt-4-1106-preview,,Proprietary,3.985
3
+ 4.175,4.14,4.1,3.98,3.789,4.235,4.06,3.788,3.414,gpt-4o-2024-05-13,,Proprietary,3.965
4
+ 4.112,4.13,3.929,4.15,4.0,4.145,4.15,3.725,3.329,gpt-4-0125-preview,,Proprietary,3.963
5
+ 4.112,4.09,3.986,3.92,3.862,4.116,4.06,3.688,3.357,gpt-4-turbo-2024-04-09,,Proprietary,3.91
6
+ 4.075,3.88,4.157,3.8,3.741,4.435,4.05,3.425,3.357,claude-3-opus-20240229,,Proprietary,3.88
7
  4.175,3.92,3.971,3.76,3.741,4.029,3.97,3.625,3.114,meta-llama/Meta-Llama-3-70B-Instruct,70.0,Chat,3.812
8
  4.075,4.03,4.0,3.83,3.776,4.13,3.96,3.325,2.771,qwen/qwen-110b-chat,110.0,Chat,3.766
9
+ 3.862,3.83,3.943,3.84,3.69,4.29,3.86,3.5,3.043,claude-3-sonnet-20240229,,Proprietary,3.762
10
+ 3.925,3.91,3.843,3.82,3.552,4.116,3.91,3.688,2.971,mistral-medium,,Proprietary,3.748
11
+ 4.0,3.94,3.957,3.58,3.569,4.275,3.93,3.538,2.871,claude-3-haiku-20240307,,Proprietary,3.74
12
+ 3.875,3.88,3.871,3.83,3.5,4.145,4.01,3.288,3.1,gemini-pro-1.5,,Proprietary,3.722
13
+ 3.9,3.83,3.757,3.66,3.638,3.957,3.94,3.712,2.871,mistral-large,,Proprietary,3.696
14
+ 4.05,3.81,3.743,3.81,3.31,4.145,3.97,3.45,2.729,google/gemini-flash-1.5,,Proprietary,3.669
15
+ 3.925,4.02,3.857,3.46,3.517,3.928,3.91,3.425,2.829,alpindale/c4ai-command-r-plus-GPTQ,104,Chat,3.652
16
+ 3.812,3.96,3.771,3.6,3.379,4.043,3.84,3.45,2.757,MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ,141,Chat,3.624
17
  3.712,3.92,3.771,3.53,3.586,4.101,3.92,3.425,2.629,Qwen/Qwen1.5-72B-Chat,72.0,Chat,3.622
18
  3.85,3.75,3.814,3.3,3.345,3.928,3.71,3.362,3.043,meta-llama/Meta-Llama-3-8B-Instruct,8.0,Chat,3.567
19
  3.775,3.86,3.8,3.44,3.534,3.986,3.91,3.325,2.429,Starling-LM-7B-beta,7.0,Chat,3.562
20
  3.65,3.85,3.643,3.55,3.121,4.246,3.8,3.488,2.671,Qwen/Qwen1.5-32B-Chat,32.0,Chat,3.558
21
  3.9,3.85,3.486,3.54,3.776,4.232,3.81,3.062,1.971,microsoft/Phi-3-mini-4k-instruct,3.8,Chat,3.514
22
  3.65,3.89,3.571,3.45,3.138,4.014,3.78,3.2,2.743,mistralai/Mixtral-8x7B-Instruct-v0.1,46.7,Chat,3.493
23
+ 3.8,3.86,3.757,3.43,3.259,3.957,3.64,2.988,2.586,gpt-3.5-turbo-0125,,Proprietary,3.475
24
+ 3.812,3.75,3.714,3.41,3.241,4.087,3.65,3.0,2.586,gpt-3.5-turbo-1106,,Proprietary,3.472
25
+ 3.562,3.65,3.629,3.48,3.069,3.884,3.74,3.062,2.986,gemini-1.0-pro,,Proprietary,3.451
26
  3.638,3.8,3.8,3.17,3.155,3.826,3.7,3.5,2.4,allenai/tulu-2-dpo-70b,70.0,Chat,3.443
27
  3.7,3.8,3.586,3.21,3.034,3.826,3.7,3.488,2.586,upstage/SOLAR-10.7B-Instruct-v1.0,10.7,Chat,3.437
28
  3.662,3.84,3.671,3.24,3.155,3.783,3.71,3.338,2.529,NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO,46.7,Chat,3.436
29
+ 3.525,3.59,3.5,3.44,3.207,3.942,3.37,2.762,,mistral-community/Mixtral-8x22B-v0.1-AWQ,141,Base,3.417
30
  3.612,3.72,3.657,2.98,3.155,4.464,3.79,2.888,2.429,meta-llama/Llama-2-70b-chat-hf,70.0,Chat,3.411
31
  3.462,3.74,3.714,3.27,3.414,4.087,3.81,2.812,2.014,01-ai/Yi-34B-Chat,34.0,Chat,3.369
32
  3.588,3.77,3.614,3.26,3.121,3.884,3.5,3.062,2.486,Qwen/Qwen1.5-14B-Chat,14.0,Chat,3.365
 
35
  3.688,3.69,3.629,3.16,3.103,3.652,3.59,3.225,2.414,NousResearch/Nous-Hermes-2-Mixtral-8x7B-SFT,46.7,Chat,3.35
36
  3.588,3.66,3.471,3.66,3.345,3.942,3.7,2.912,1.814,microsoft/Phi-3-mini-128k-instruct,3.8,Chat,3.344
37
  3.525,3.76,3.514,3.26,3.31,3.841,3.61,2.888,2.314,openchat/openchat-3.5-0106,7.0,Chat,3.336
38
+ 3.288,3.62,3.686,3.25,3.345,3.551,3.45,3.062,2.543,MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ,141,Chat,3.31
39
  3.712,3.74,3.5,3.2,2.948,3.942,3.53,2.838,2.129,Starling-LM-7B-alpha,7.0,Chat,3.282
40
  3.4,3.74,3.4,3.04,3.0,3.754,3.71,2.975,2.043,Qwen/Qwen1.5-7B-Chat,7.0,Chat,3.229
41
  3.588,3.7,3.343,2.71,2.862,4.319,3.66,2.512,2.343,meta-llama/Llama-2-13b-chat-hf,13.0,Chat,3.226
data/bgb-leaderboard-prometheus-bgb-8x7b-v2.0.pkl CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e542c1e3dbe78dfd4bcfaac73c0a8b27e1a8f117a323b9a7b08600bb4c292034
3
- size 13924
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:54f03922163c1c21e91e3c3e11d29b02c7218be0b5547c6461e0a1d49f8bd68c
3
+ size 13905
requirements.txt CHANGED
@@ -1,5 +1,4 @@
1
  huggingface_hub
2
- transformers
3
  gradio
4
  plotly
5
  pandas
 
1
  huggingface_hub
 
2
  gradio
3
  plotly
4
  pandas
src/llm_perf.py CHANGED
@@ -162,10 +162,10 @@ def get_eval_df(eval_model_name: str):
162
  eval_df.drop(columns=["model_name"], inplace=True)
163
 
164
  eval_df["model_params"] = eval_df["model_name_or_path"].apply(
165
- lambda x: MODEL_MAPPING.get(x, ["Unknown", "Unknown"])[0]
166
  )
167
  eval_df["model_type"] = eval_df["model_name_or_path"].apply(
168
- lambda x: MODEL_MAPPING.get(x, ["Unknown", "Unknown"])[1]
169
  )
170
 
171
  capabilities = [
 
162
  eval_df.drop(columns=["model_name"], inplace=True)
163
 
164
  eval_df["model_params"] = eval_df["model_name_or_path"].apply(
165
+ lambda x: MODEL_MAPPING.get(x, [None, "Unknown"])[0]
166
  )
167
  eval_df["model_type"] = eval_df["model_name_or_path"].apply(
168
+ lambda x: MODEL_MAPPING.get(x, [None, "Unknown"])[1]
169
  )
170
 
171
  capabilities = [
src/model_list.py CHANGED
@@ -391,32 +391,32 @@ MODEL_MAPPING = {
391
  "CohereForAI/c4ai-command-r-v01": [35.0, "Chat"],
392
  "meta-llama/Llama-2-70b-hf": [70.0, "Base"],
393
  "codellama/CodeLlama-70b-hf": [70.0, "Base"],
394
- "mistral-community/Mixtral-8x22B-v0.1-AWQ": ["AWQ", "Base"],
395
  "meta-llama/Meta-Llama-3-70B": [70.0, "Base"],
396
  "Qwen/Qwen1.5-72B": [72.0, "Base"],
397
  "meta-llama/Llama-2-70b-chat-hf": [70.0, "Chat", "meta-llama/Llama-2-70b-hf"],
398
  "codellama/CodeLlama-70b-Instruct-hf": [70.0, "Chat", "codellama/CodeLlama-70b-hf"],
399
  "allenai/tulu-2-dpo-70b": [70.0, "Chat", "meta-llama/Llama-2-70b-hf"],
400
- "alpindale/c4ai-command-r-plus-GPTQ": ["GPTQ", "Chat"],
401
  "meta-llama/Meta-Llama-3-70B-Instruct": [70.0, "Chat", "meta-llama/Meta-Llama-3-70B"],
402
- "MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ": ["AWQ", "Chat", "mistral-community/Mixtral-8x22B-v0.1-AWQ"],
403
- "MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ": ["AWQ", "Chat", "mistral-community/Mixtral-8x22B-v0.1-AWQ"],
404
  "Qwen/Qwen1.5-72B-Chat": [72.0, "Chat", "Qwen/Qwen1.5-72B"],
405
  "qwen/qwen-110b-chat": [110.0, "Chat", None],
406
- "gpt-3.5-turbo-1106": ["Proprietary", "Proprietary"],
407
- "gpt-3.5-turbo-0125": ["Proprietary", "Proprietary"],
408
- "gpt-4-1106-preview": ["Proprietary", "Proprietary"],
409
- "gpt-4-0125-preview": ["Proprietary", "Proprietary"],
410
- "gpt-4-turbo-2024-04-09": ["Proprietary", "Proprietary"],
411
- "gpt-4o-2024-05-13": ["Proprietary", "Proprietary"],
412
- "mistral-medium": ["Proprietary", "Proprietary"],
413
- "mistral-large": ["Proprietary", "Proprietary"],
414
- "gemini-1.0-pro": ["Proprietary", "Proprietary"],
415
- "gemini-pro-1.5": ["Proprietary", "Proprietary"],
416
- "google/gemini-flash-1.5": ["Proprietary", "Proprietary"],
417
- "claude-3-haiku-20240307": ["Proprietary", "Proprietary"],
418
- "claude-3-sonnet-20240229": ["Proprietary", "Proprietary"],
419
- "claude-3-opus-20240229": ["Proprietary", "Proprietary"],
420
  }
421
 
422
 
 
391
  "CohereForAI/c4ai-command-r-v01": [35.0, "Chat"],
392
  "meta-llama/Llama-2-70b-hf": [70.0, "Base"],
393
  "codellama/CodeLlama-70b-hf": [70.0, "Base"],
394
+ "mistral-community/Mixtral-8x22B-v0.1-AWQ": [141, "Base"],
395
  "meta-llama/Meta-Llama-3-70B": [70.0, "Base"],
396
  "Qwen/Qwen1.5-72B": [72.0, "Base"],
397
  "meta-llama/Llama-2-70b-chat-hf": [70.0, "Chat", "meta-llama/Llama-2-70b-hf"],
398
  "codellama/CodeLlama-70b-Instruct-hf": [70.0, "Chat", "codellama/CodeLlama-70b-hf"],
399
  "allenai/tulu-2-dpo-70b": [70.0, "Chat", "meta-llama/Llama-2-70b-hf"],
400
+ "alpindale/c4ai-command-r-plus-GPTQ": [104, "Chat"],
401
  "meta-llama/Meta-Llama-3-70B-Instruct": [70.0, "Chat", "meta-llama/Meta-Llama-3-70B"],
402
+ "MaziyarPanahi/Mixtral-8x22B-Instruct-v0.1-AWQ": ["141", "Chat", "mistral-community/Mixtral-8x22B-v0.1-AWQ"],
403
+ "MaziyarPanahi/zephyr-orpo-141b-A35b-v0.1-AWQ": ["141", "Chat", "mistral-community/Mixtral-8x22B-v0.1-AWQ"],
404
  "Qwen/Qwen1.5-72B-Chat": [72.0, "Chat", "Qwen/Qwen1.5-72B"],
405
  "qwen/qwen-110b-chat": [110.0, "Chat", None],
406
+ "gpt-3.5-turbo-1106": [None, "Proprietary"],
407
+ "gpt-3.5-turbo-0125": [None, "Proprietary"],
408
+ "gpt-4-1106-preview": [None, "Proprietary"],
409
+ "gpt-4-0125-preview": [None, "Proprietary"],
410
+ "gpt-4-turbo-2024-04-09": [None, "Proprietary"],
411
+ "gpt-4o-2024-05-13": [None, "Proprietary"],
412
+ "mistral-medium": [None, "Proprietary"],
413
+ "mistral-large": [None, "Proprietary"],
414
+ "gemini-1.0-pro": [None, "Proprietary"],
415
+ "gemini-pro-1.5": [None, "Proprietary"],
416
+ "google/gemini-flash-1.5": [None, "Proprietary"],
417
+ "claude-3-haiku-20240307": [None, "Proprietary"],
418
+ "claude-3-sonnet-20240229": [None, "Proprietary"],
419
+ "claude-3-opus-20240229": [None, "Proprietary"],
420
  }
421
 
422