Files changed (1) hide show
  1. README.md +112 -4
README.md CHANGED
@@ -1,13 +1,108 @@
1
  ---
 
2
  library_name: transformers
3
  tags:
4
  - trl
5
  - sft
6
- license: apache-2.0
7
- datasets:
8
- - HuggingFaceTB/smoltalk
9
  base_model:
10
  - nbeerbower/Mahou-1.5-mistral-nemo-12B-lorablated
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
11
  ---
12
 
13
  ![image/png](https://huggingface.co/nbeerbower/SmolNemo-12B/resolve/main/smolnemo_cover.png?download=true)
@@ -26,4 +121,17 @@ base_model:
26
 
27
  SFT with 8x A100 for 0.1 epochs.
28
 
29
- This was a full finetune. I think the issues with the model can be chalked up to conflicts with Mistral Instruct and ChatML.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ license: apache-2.0
3
  library_name: transformers
4
  tags:
5
  - trl
6
  - sft
 
 
 
7
  base_model:
8
  - nbeerbower/Mahou-1.5-mistral-nemo-12B-lorablated
9
+ datasets:
10
+ - HuggingFaceTB/smoltalk
11
+ model-index:
12
+ - name: SmolNemo-12B-FFT-experimental
13
+ results:
14
+ - task:
15
+ type: text-generation
16
+ name: Text Generation
17
+ dataset:
18
+ name: IFEval (0-Shot)
19
+ type: HuggingFaceH4/ifeval
20
+ args:
21
+ num_few_shot: 0
22
+ metrics:
23
+ - type: inst_level_strict_acc and prompt_level_strict_acc
24
+ value: 33.48
25
+ name: strict accuracy
26
+ source:
27
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
28
+ name: Open LLM Leaderboard
29
+ - task:
30
+ type: text-generation
31
+ name: Text Generation
32
+ dataset:
33
+ name: BBH (3-Shot)
34
+ type: BBH
35
+ args:
36
+ num_few_shot: 3
37
+ metrics:
38
+ - type: acc_norm
39
+ value: 6.54
40
+ name: normalized accuracy
41
+ source:
42
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
43
+ name: Open LLM Leaderboard
44
+ - task:
45
+ type: text-generation
46
+ name: Text Generation
47
+ dataset:
48
+ name: MATH Lvl 5 (4-Shot)
49
+ type: hendrycks/competition_math
50
+ args:
51
+ num_few_shot: 4
52
+ metrics:
53
+ - type: exact_match
54
+ value: 0.23
55
+ name: exact match
56
+ source:
57
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
58
+ name: Open LLM Leaderboard
59
+ - task:
60
+ type: text-generation
61
+ name: Text Generation
62
+ dataset:
63
+ name: GPQA (0-shot)
64
+ type: Idavidrein/gpqa
65
+ args:
66
+ num_few_shot: 0
67
+ metrics:
68
+ - type: acc_norm
69
+ value: 1.34
70
+ name: acc_norm
71
+ source:
72
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
73
+ name: Open LLM Leaderboard
74
+ - task:
75
+ type: text-generation
76
+ name: Text Generation
77
+ dataset:
78
+ name: MuSR (0-shot)
79
+ type: TAUR-Lab/MuSR
80
+ args:
81
+ num_few_shot: 0
82
+ metrics:
83
+ - type: acc_norm
84
+ value: 5.92
85
+ name: acc_norm
86
+ source:
87
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
88
+ name: Open LLM Leaderboard
89
+ - task:
90
+ type: text-generation
91
+ name: Text Generation
92
+ dataset:
93
+ name: MMLU-PRO (5-shot)
94
+ type: TIGER-Lab/MMLU-Pro
95
+ config: main
96
+ split: test
97
+ args:
98
+ num_few_shot: 5
99
+ metrics:
100
+ - type: acc
101
+ value: 2.41
102
+ name: accuracy
103
+ source:
104
+ url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=nbeerbower/SmolNemo-12B-FFT-experimental
105
+ name: Open LLM Leaderboard
106
  ---
107
 
108
  ![image/png](https://huggingface.co/nbeerbower/SmolNemo-12B/resolve/main/smolnemo_cover.png?download=true)
 
121
 
122
  SFT with 8x A100 for 0.1 epochs.
123
 
124
+ This was a full finetune. I think the issues with the model can be chalked up to conflicts with Mistral Instruct and ChatML.
125
+ # [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
126
+ Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_nbeerbower__SmolNemo-12B-FFT-experimental)
127
+
128
+ | Metric |Value|
129
+ |-------------------|----:|
130
+ |Avg. | 8.32|
131
+ |IFEval (0-Shot) |33.48|
132
+ |BBH (3-Shot) | 6.54|
133
+ |MATH Lvl 5 (4-Shot)| 0.23|
134
+ |GPQA (0-shot) | 1.34|
135
+ |MuSR (0-shot) | 5.92|
136
+ |MMLU-PRO (5-shot) | 2.41|
137
+