Adding Evaluation Results (#1)
Browse files- Adding Evaluation Results (f28cf641514cfeebc3145d8209ee940e9f5a1802)
Co-authored-by: Open LLM Leaderboard PR Bot <[email protected]>
README.md
CHANGED
|
@@ -1,76 +1,233 @@
|
|
| 1 |
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
language:
|
| 4 |
-
|
|
|
|
| 5 |
tags:
|
| 6 |
-
|
| 7 |
base_model: BEE-spoke-data/smol_llama-101M-GQA
|
| 8 |
datasets:
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
widget:
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
inference:
|
| 69 |
parameters:
|
| 70 |
max_new_tokens: 250
|
| 71 |
penalty_alpha: 0.5
|
| 72 |
top_k: 4
|
| 73 |
repetition_penalty: 1.105
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 74 |
---
|
| 75 |
|
| 76 |
# A Llama Chat Model of 101M Parameters
|
|
@@ -105,3 +262,17 @@ penalty_alpha: 0.5
|
|
| 105 |
top_k: 4
|
| 106 |
repetition_penalty: 1.105
|
| 107 |
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
|
|
|
| 2 |
language:
|
| 3 |
+
- en
|
| 4 |
+
license: apache-2.0
|
| 5 |
tags:
|
| 6 |
+
- text-generation
|
| 7 |
base_model: BEE-spoke-data/smol_llama-101M-GQA
|
| 8 |
datasets:
|
| 9 |
+
- Open-Orca/SlimOrca-Dedup
|
| 10 |
+
- VMware/open-instruct
|
| 11 |
+
- LDJnr/Capybara
|
| 12 |
+
- cognitivecomputations/ultrachat-uncensored
|
| 13 |
+
- starfishmedical/webGPT_x_dolly
|
| 14 |
+
- THUDM/webglm-qa
|
| 15 |
widget:
|
| 16 |
+
- text: '<|im_start|>system
|
| 17 |
+
|
| 18 |
+
You are a helpful assistant who gives creative responses.<|im_end|>
|
| 19 |
+
|
| 20 |
+
<|im_start|>user
|
| 21 |
+
|
| 22 |
+
Write the background story of a game about wizards and llamas in a sci-fi world.<|im_end|>
|
| 23 |
+
|
| 24 |
+
<|im_start|>assistant'
|
| 25 |
+
- text: '<|im_start|>system
|
| 26 |
+
|
| 27 |
+
A friendly chat between a user and an assistant.<|im_end|>
|
| 28 |
+
|
| 29 |
+
<|im_start|>user
|
| 30 |
+
|
| 31 |
+
Got a question for you!<|im_end|>
|
| 32 |
+
|
| 33 |
+
<|im_start|>assistant
|
| 34 |
+
|
| 35 |
+
Sure! What''s it?<|im_end|>
|
| 36 |
+
|
| 37 |
+
<|im_start|>user
|
| 38 |
+
|
| 39 |
+
I need to build a simple website. Where should I start learning about web development?<|im_end|>
|
| 40 |
+
|
| 41 |
+
<|im_start|>assistant'
|
| 42 |
+
- text: '<|im_start|>system
|
| 43 |
+
|
| 44 |
+
You are a helpful assistant who provides concise answers to the user''s questions.<|im_end|>
|
| 45 |
+
|
| 46 |
+
<|im_start|>user
|
| 47 |
+
|
| 48 |
+
How to become more healthy?<|im_end|>
|
| 49 |
+
|
| 50 |
+
<|im_start|>assistant'
|
| 51 |
+
- text: '<|im_start|>system
|
| 52 |
+
|
| 53 |
+
You are a helpful assistant, who always answers with empathy.<|im_end|>
|
| 54 |
+
|
| 55 |
+
<|im_start|>user
|
| 56 |
+
|
| 57 |
+
List the pros and cons of social media.<|im_end|>
|
| 58 |
+
|
| 59 |
+
<|im_start|>assistant'
|
| 60 |
+
- text: '<|im_start|>system
|
| 61 |
+
|
| 62 |
+
You are a helpful assistant, who always answers with empathy.<|im_end|>
|
| 63 |
+
|
| 64 |
+
<|im_start|>user
|
| 65 |
+
|
| 66 |
+
Hello!<|im_end|>
|
| 67 |
+
|
| 68 |
+
<|im_start|>assistant
|
| 69 |
+
|
| 70 |
+
Hi! How can I help you today?<|im_end|>
|
| 71 |
+
|
| 72 |
+
<|im_start|>user
|
| 73 |
+
|
| 74 |
+
Take a look at the info below.
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
- The tape inside the VHS cassettes is very delicate and can be easily ruined,
|
| 78 |
+
making them unplayable and unrepairable. The reason the tape deteriorates is that
|
| 79 |
+
the magnetic charge needed for them to work is not permanent, and the magnetic
|
| 80 |
+
particles end up losing their charge in a process known as remanence decay. These
|
| 81 |
+
particles could also become demagnetised via being stored too close to a magnetic
|
| 82 |
+
source.
|
| 83 |
+
|
| 84 |
+
- One of the most significant issues with VHS tapes is that they have moving parts,
|
| 85 |
+
meaning that there are more occasions when something can go wrong, damaging your
|
| 86 |
+
footage or preventing it from playing back. The tape itself is a prominent cause
|
| 87 |
+
of this, and tape slippage can occur. Tapes slippage can be caused when the tape
|
| 88 |
+
loses its tension, or it has become warped. These problems can occur in storage
|
| 89 |
+
due to high temperatures or frequent changes in humidity.
|
| 90 |
+
|
| 91 |
+
- VHS tapes deteriorate over time from infrequent or overuse. Neglect means mold
|
| 92 |
+
and dirt, while overuse can lead to scratches and technical difficulties. This
|
| 93 |
+
is why old VHS tapes inevitably experience malfunctions after a long period of
|
| 94 |
+
time. Usually anywhere between 10 to 25+ years.
|
| 95 |
+
|
| 96 |
+
- Some VHS tapes like newer mini DVs and Digital 8 tapes can suffer from digital
|
| 97 |
+
corruption, meaning that the footage becomes lost and cannot be recovered. These
|
| 98 |
+
tapes were the steppingstone from VHS to the digital age when capturing footage
|
| 99 |
+
straight to digital became the norm. Unfortunately,they are susceptible to digital
|
| 100 |
+
corruption, which causes video pixilation and/or loss of audio.<|im_end|>
|
| 101 |
+
|
| 102 |
+
<|im_start|>assistant
|
| 103 |
+
|
| 104 |
+
Alright!<|im_end|>
|
| 105 |
+
|
| 106 |
+
<|im_start|>user
|
| 107 |
+
|
| 108 |
+
Now I''m going to write my question, and if the info above is useful, you can
|
| 109 |
+
use them in your response.
|
| 110 |
+
|
| 111 |
+
Ready?<|im_end|>
|
| 112 |
+
|
| 113 |
+
<|im_start|>assistant
|
| 114 |
+
|
| 115 |
+
Ready for your question!<|im_end|>
|
| 116 |
+
|
| 117 |
+
<|im_start|>user
|
| 118 |
+
|
| 119 |
+
Why do VHS tapes deteriorate over time?<|im_end|>
|
| 120 |
+
|
| 121 |
+
<|im_start|>assistant'
|
| 122 |
inference:
|
| 123 |
parameters:
|
| 124 |
max_new_tokens: 250
|
| 125 |
penalty_alpha: 0.5
|
| 126 |
top_k: 4
|
| 127 |
repetition_penalty: 1.105
|
| 128 |
+
model-index:
|
| 129 |
+
- name: Smol-Llama-101M-Chat-v1
|
| 130 |
+
results:
|
| 131 |
+
- task:
|
| 132 |
+
type: text-generation
|
| 133 |
+
name: Text Generation
|
| 134 |
+
dataset:
|
| 135 |
+
name: AI2 Reasoning Challenge (25-Shot)
|
| 136 |
+
type: ai2_arc
|
| 137 |
+
config: ARC-Challenge
|
| 138 |
+
split: test
|
| 139 |
+
args:
|
| 140 |
+
num_few_shot: 25
|
| 141 |
+
metrics:
|
| 142 |
+
- type: acc_norm
|
| 143 |
+
value: 22.87
|
| 144 |
+
name: normalized accuracy
|
| 145 |
+
source:
|
| 146 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 147 |
+
name: Open LLM Leaderboard
|
| 148 |
+
- task:
|
| 149 |
+
type: text-generation
|
| 150 |
+
name: Text Generation
|
| 151 |
+
dataset:
|
| 152 |
+
name: HellaSwag (10-Shot)
|
| 153 |
+
type: hellaswag
|
| 154 |
+
split: validation
|
| 155 |
+
args:
|
| 156 |
+
num_few_shot: 10
|
| 157 |
+
metrics:
|
| 158 |
+
- type: acc_norm
|
| 159 |
+
value: 28.69
|
| 160 |
+
name: normalized accuracy
|
| 161 |
+
source:
|
| 162 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 163 |
+
name: Open LLM Leaderboard
|
| 164 |
+
- task:
|
| 165 |
+
type: text-generation
|
| 166 |
+
name: Text Generation
|
| 167 |
+
dataset:
|
| 168 |
+
name: MMLU (5-Shot)
|
| 169 |
+
type: cais/mmlu
|
| 170 |
+
config: all
|
| 171 |
+
split: test
|
| 172 |
+
args:
|
| 173 |
+
num_few_shot: 5
|
| 174 |
+
metrics:
|
| 175 |
+
- type: acc
|
| 176 |
+
value: 24.93
|
| 177 |
+
name: accuracy
|
| 178 |
+
source:
|
| 179 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 180 |
+
name: Open LLM Leaderboard
|
| 181 |
+
- task:
|
| 182 |
+
type: text-generation
|
| 183 |
+
name: Text Generation
|
| 184 |
+
dataset:
|
| 185 |
+
name: TruthfulQA (0-shot)
|
| 186 |
+
type: truthful_qa
|
| 187 |
+
config: multiple_choice
|
| 188 |
+
split: validation
|
| 189 |
+
args:
|
| 190 |
+
num_few_shot: 0
|
| 191 |
+
metrics:
|
| 192 |
+
- type: mc2
|
| 193 |
+
value: 45.76
|
| 194 |
+
source:
|
| 195 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 196 |
+
name: Open LLM Leaderboard
|
| 197 |
+
- task:
|
| 198 |
+
type: text-generation
|
| 199 |
+
name: Text Generation
|
| 200 |
+
dataset:
|
| 201 |
+
name: Winogrande (5-shot)
|
| 202 |
+
type: winogrande
|
| 203 |
+
config: winogrande_xl
|
| 204 |
+
split: validation
|
| 205 |
+
args:
|
| 206 |
+
num_few_shot: 5
|
| 207 |
+
metrics:
|
| 208 |
+
- type: acc
|
| 209 |
+
value: 50.04
|
| 210 |
+
name: accuracy
|
| 211 |
+
source:
|
| 212 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 213 |
+
name: Open LLM Leaderboard
|
| 214 |
+
- task:
|
| 215 |
+
type: text-generation
|
| 216 |
+
name: Text Generation
|
| 217 |
+
dataset:
|
| 218 |
+
name: GSM8k (5-shot)
|
| 219 |
+
type: gsm8k
|
| 220 |
+
config: main
|
| 221 |
+
split: test
|
| 222 |
+
args:
|
| 223 |
+
num_few_shot: 5
|
| 224 |
+
metrics:
|
| 225 |
+
- type: acc
|
| 226 |
+
value: 0.08
|
| 227 |
+
name: accuracy
|
| 228 |
+
source:
|
| 229 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=Felladrin/Smol-Llama-101M-Chat-v1
|
| 230 |
+
name: Open LLM Leaderboard
|
| 231 |
---
|
| 232 |
|
| 233 |
# A Llama Chat Model of 101M Parameters
|
|
|
|
| 262 |
top_k: 4
|
| 263 |
repetition_penalty: 1.105
|
| 264 |
```
|
| 265 |
+
|
| 266 |
+
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
| 267 |
+
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_Felladrin__Smol-Llama-101M-Chat-v1)
|
| 268 |
+
|
| 269 |
+
| Metric |Value|
|
| 270 |
+
|---------------------------------|----:|
|
| 271 |
+
|Avg. |28.73|
|
| 272 |
+
|AI2 Reasoning Challenge (25-Shot)|22.87|
|
| 273 |
+
|HellaSwag (10-Shot) |28.69|
|
| 274 |
+
|MMLU (5-Shot) |24.93|
|
| 275 |
+
|TruthfulQA (0-shot) |45.76|
|
| 276 |
+
|Winogrande (5-shot) |50.04|
|
| 277 |
+
|GSM8k (5-shot) | 0.08|
|
| 278 |
+
|