Weyaxi commited on
Commit
ba4cb82
1 Parent(s): 97c24b1

Adding Evaluation Results

Browse files

This is an automated PR created with https://huggingface.co/spaces/Weyaxi/open-llm-leaderboard-results-pr

The purpose of this PR is to add evaluation results from the Open LLM Leaderboard to your model card.

If you encounter any issues, please report them to https://huggingface.co/spaces/Weyaxi/open-llm-leaderboard-results-pr/discussions

Files changed (1) hide show
  1. README.md +117 -1
README.md CHANGED
@@ -1,5 +1,108 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
4
  this is [miqu-1-70b](https://huggingface.co/miqudev/miqu-1-70b), dequantised from q5 to f16 && transposed to pytorch. shapes have been rotated less wrongly than in [alpindale/miqu-1-70b-pytorch](https://huggingface.co/alpindale/miqu-1-70b-pytorch/tree/main)
5
 
@@ -130,4 +233,17 @@ some benchmarks
130
  ```
131
  no i do not know why the stderr is high. plausibly it is due to the vllm backend used. this is my lm-eval command in most cases (works on h100):
132
 
133
- `lm_eval --model vllm --model_args pretrained=./miqu-1-70b-sf,tensor_parallel_size=4,dtype=auto,gpu_memory_utilization=0.88,data_parallel_size=2 --tasks mmlu --batch_size 20`
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
+ model-index:
4
+ - name: miqu-1-70b-sf
5
+ results:
6
+ - task:
7
+ type: text-generation
8
+ name: Text Generation
9
+ dataset:
10
+ name: AI2 Reasoning Challenge (25-Shot)
11
+ type: ai2_arc
12
+ config: ARC-Challenge
13
+ split: test
14
+ args:
15
+ num_few_shot: 25
16
+ metrics:
17
+ - type: acc_norm
18
+ value: 73.04
19
+ name: normalized accuracy
20
+ source:
21
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
22
+ name: Open LLM Leaderboard
23
+ - task:
24
+ type: text-generation
25
+ name: Text Generation
26
+ dataset:
27
+ name: HellaSwag (10-Shot)
28
+ type: hellaswag
29
+ split: validation
30
+ args:
31
+ num_few_shot: 10
32
+ metrics:
33
+ - type: acc_norm
34
+ value: 88.61
35
+ name: normalized accuracy
36
+ source:
37
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
38
+ name: Open LLM Leaderboard
39
+ - task:
40
+ type: text-generation
41
+ name: Text Generation
42
+ dataset:
43
+ name: MMLU (5-Shot)
44
+ type: cais/mmlu
45
+ config: all
46
+ split: test
47
+ args:
48
+ num_few_shot: 5
49
+ metrics:
50
+ - type: acc
51
+ value: 75.49
52
+ name: accuracy
53
+ source:
54
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
55
+ name: Open LLM Leaderboard
56
+ - task:
57
+ type: text-generation
58
+ name: Text Generation
59
+ dataset:
60
+ name: TruthfulQA (0-shot)
61
+ type: truthful_qa
62
+ config: multiple_choice
63
+ split: validation
64
+ args:
65
+ num_few_shot: 0
66
+ metrics:
67
+ - type: mc2
68
+ value: 69.38
69
+ source:
70
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
71
+ name: Open LLM Leaderboard
72
+ - task:
73
+ type: text-generation
74
+ name: Text Generation
75
+ dataset:
76
+ name: Winogrande (5-shot)
77
+ type: winogrande
78
+ config: winogrande_xl
79
+ split: validation
80
+ args:
81
+ num_few_shot: 5
82
+ metrics:
83
+ - type: acc
84
+ value: 85.32
85
+ name: accuracy
86
+ source:
87
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
88
+ name: Open LLM Leaderboard
89
+ - task:
90
+ type: text-generation
91
+ name: Text Generation
92
+ dataset:
93
+ name: GSM8k (5-shot)
94
+ type: gsm8k
95
+ config: main
96
+ split: test
97
+ args:
98
+ num_few_shot: 5
99
+ metrics:
100
+ - type: acc
101
+ value: 67.7
102
+ name: accuracy
103
+ source:
104
+ url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=152334H/miqu-1-70b-sf
105
+ name: Open LLM Leaderboard
106
  ---
107
  this is [miqu-1-70b](https://huggingface.co/miqudev/miqu-1-70b), dequantised from q5 to f16 && transposed to pytorch. shapes have been rotated less wrongly than in [alpindale/miqu-1-70b-pytorch](https://huggingface.co/alpindale/miqu-1-70b-pytorch/tree/main)
108
 
 
233
  ```
234
  no i do not know why the stderr is high. plausibly it is due to the vllm backend used. this is my lm-eval command in most cases (works on h100):
235
 
236
+ `lm_eval --model vllm --model_args pretrained=./miqu-1-70b-sf,tensor_parallel_size=4,dtype=auto,gpu_memory_utilization=0.88,data_parallel_size=2 --tasks mmlu --batch_size 20`
237
+ # [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
238
+ Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_152334H__miqu-1-70b-sf)
239
+
240
+ | Metric |Value|
241
+ |---------------------------------|----:|
242
+ |Avg. |76.59|
243
+ |AI2 Reasoning Challenge (25-Shot)|73.04|
244
+ |HellaSwag (10-Shot) |88.61|
245
+ |MMLU (5-Shot) |75.49|
246
+ |TruthfulQA (0-shot) |69.38|
247
+ |Winogrande (5-shot) |85.32|
248
+ |GSM8k (5-shot) |67.70|
249
+