Hugging Face's logo

SmallDoge
/
Doge-160M-Instruct

Question Answering
Transformers
Safetensors
English
doge
text-generation
trl
sft
dpo
custom_code
Model card Files
xet
Community
JingzeShi commited on
Commit
2efe10b
路
verified 路
1 Parent(s): babd934

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -2
README.md CHANGED
@@ -90,7 +90,9 @@ We build the Doge-Instruct by first SFT on [SmolTalk](https://huggingface.co/dat
90
  | Model | Training Data | Epochs | Content Length | LR | Batch Size | Precision |
91
  |---|---|---|---|---|---|---|
92
  | [Doge-20M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-20M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 8e-4 | 0.25M | bfloat16 |
 
93
  | [Doge-60M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-60M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 6e-4 | 0.25M | bfloat16 |
 
94
  | [Doge-160M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-160M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 4e-4 | 0.25M | bfloat16 |
95
  | [Doge-320M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-320M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 2e-4 | 0.25M | bfloat16 |
96
 
@@ -98,7 +100,9 @@ We build the Doge-Instruct by first SFT on [SmolTalk](https://huggingface.co/dat
98
  | Model | Training Data | Epochs | Content Length | LR | Batch Size | Precision |
99
  |---|---|---|---|---|---|---|
100
  | [Doge-20M-Instruct](https://huggingface.co/SmallDoge/Doge-20M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 8e-5 | 0.125M | bfloat16 |
 
101
  | [Doge-60M-Instruct](https://huggingface.co/SmallDoge/Doge-60M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 6e-5 | 0.125M | bfloat16 |
 
102
  | [Doge-160M-Instruct](https://huggingface.co/SmallDoge/Doge-160M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 4e-5 | 0.125M | bfloat16 |
103
  | [Doge-320M-Instruct](https://huggingface.co/SmallDoge/Doge-320M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 2e-5 | 0.125M | bfloat16 |
104
 
@@ -107,8 +111,10 @@ We build the Doge-Instruct by first SFT on [SmolTalk](https://huggingface.co/dat
107
 
108
  | Model | IFEval (Prompt Strict Acc) | MMLU | BBH | ARC | PIQA | HellaSwag | tokens / s on i7-11 CPU |
109
  |---|---|---|---|---|---|---|---|
110
- | [Doge-20M-Instruct](https://huggingface.co/SmallDoge/Doge-20M-Instruct) | 7.3 | 26.3 | 18.3 | 29.2 | 57.8 | 27.8 | 142 |
111
- | [Doge-60M-Instruct](https://huggingface.co/SmallDoge/Doge-60M-Instruct) | 7.4 | 27.5 | 27.7 | 37.5 | 61.4 | 32.1 | 62 |
 
 
112
  | [Doge-160M-Instruct](https://huggingface.co/SmallDoge/Doge-160M-Instruct) | 16.8 | 29.7 | 29.1 | 42.8 | 64.1 | 37.1 | 28 |
113
  | [Doge-320M-Instruct](https://huggingface.co/SmallDoge/Doge-320M-Instruct) | 28.5 | 30.3 | 31.9 | 51.7 | 71.0 | 50.6 | 16 |
114
 
 
90
  | Model | Training Data | Epochs | Content Length | LR | Batch Size | Precision |
91
  |---|---|---|---|---|---|---|
92
  | [Doge-20M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-20M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 8e-4 | 0.25M | bfloat16 |
93
+ | [Doge-20M-MoE-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-20M-MoE-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 8e-4 | 0.25M | bfloat16 |
94
  | [Doge-60M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-60M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 6e-4 | 0.25M | bfloat16 |
95
+ | [Doge-120M-MoE-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-120M-MoE-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 6e-4 | 0.25M | bfloat16 |
96
  | [Doge-160M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-160M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 4e-4 | 0.25M | bfloat16 |
97
  | [Doge-320M-Instruct-SFT](https://huggingface.co/SmallDoge/Doge-320M-Instruct-SFT) | [smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk) | 2 | 2048 | 2e-4 | 0.25M | bfloat16 |
98
 
 
100
  | Model | Training Data | Epochs | Content Length | LR | Batch Size | Precision |
101
  |---|---|---|---|---|---|---|
102
  | [Doge-20M-Instruct](https://huggingface.co/SmallDoge/Doge-20M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 8e-5 | 0.125M | bfloat16 |
103
+ | [Doge-20M-MoE-Instruct](https://huggingface.co/SmallDoge/Doge-20M-MoE-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 8e-5 | 0.125M | bfloat16 |
104
  | [Doge-60M-Instruct](https://huggingface.co/SmallDoge/Doge-60M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 6e-5 | 0.125M | bfloat16 |
105
+ | [Doge-120M-MoE-Instruct](https://huggingface.co/SmallDoge/Doge-120M-MoE-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 6e-5 | 0.125M | bfloat16 |
106
  | [Doge-160M-Instruct](https://huggingface.co/SmallDoge/Doge-160M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 4e-5 | 0.125M | bfloat16 |
107
  | [Doge-320M-Instruct](https://huggingface.co/SmallDoge/Doge-320M-Instruct) | [ultrafeedback_binarized](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) | 2 | 1024 | 2e-5 | 0.125M | bfloat16 |
108
 
 
111
 
112
  | Model | IFEval (Prompt Strict Acc) | MMLU | BBH | ARC | PIQA | HellaSwag | tokens / s on i7-11 CPU |
113
  |---|---|---|---|---|---|---|---|
114
+ | [Doge-20M-Instruct](https://huggingface.co/SmallDoge/Doge-20M-Instruct) | 9.2 | 26.3 | 18.3 | 29.2 | 57.8 | 27.8 | 142 |
115
+ | [Doge-20M-MoE-Instruct](https://huggingface.co/SmallDoge/Doge-20M-MoE-Instruct) | 13.7 | 26.5 | 26.3 | 31.1 | 58.2 | 27.9 | 132 |
116
+ | [Doge-60M-Instruct](https://huggingface.co/SmallDoge/Doge-60M-Instruct) | 9.4 | 27.5 | 27.7 | 37.5 | 61.4 | 32.1 | 62 |
117
+ | [Doge-120M-MoE-Instruct](https://huggingface.co/SmallDoge/Doge-120M-MoE-Instruct) | 24.4 | 28.2 | 30.1 | 44.2 | 62.1 | 36.3 | 58 |
118
  | [Doge-160M-Instruct](https://huggingface.co/SmallDoge/Doge-160M-Instruct) | 16.8 | 29.7 | 29.1 | 42.8 | 64.1 | 37.1 | 28 |
119
  | [Doge-320M-Instruct](https://huggingface.co/SmallDoge/Doge-320M-Instruct) | 28.5 | 30.3 | 31.9 | 51.7 | 71.0 | 50.6 | 16 |
120