File size: 5,756 Bytes
f774aea
 
 
 
4a9b9a9
f774aea
 
 
 
 
 
 
 
 
356848c
364d895
356848c
f774aea
4a9b9a9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f774aea
 
 
 
 
c7c0dba
 
f774aea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ca26e2b
 
 
 
 
abc0285
ca26e2b
 
 
0e7d5ab
 
 
 
 
 
 
 
 
 
 
 
 
356848c
364d895
356848c
0e7d5ab
 
 
 
 
 
 
 
 
 
4a9b9a9
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
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
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
---
license: apache-2.0
datasets:
- teknium/openhermes
base_model: BEE-spoke-data/smol_llama-220M-GQA
inference:
  parameters:
    do_sample: true
    renormalize_logits: true
    temperature: 0.25
    top_p: 0.95
    top_k: 50
    min_new_tokens: 2
    max_new_tokens: 96
    repetition_penalty: 1.03
    no_repeat_ngram_size: 5
    epsilon_cutoff: 0.0008
widget:
- text: "Below is an instruction that describes a task, paired with an input that\
    \ provides further context. Write a response that appropriately completes the\
    \ request.  \n   \n### Instruction:  \n  \nWrite an ode to Chipotle burritos.\
    \ \n  \n### Response:  \n"
  example_title: burritos
model-index:
- name: smol_llama-220M-openhermes
  results:
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: AI2 Reasoning Challenge (25-Shot)
      type: ai2_arc
      config: ARC-Challenge
      split: test
      args:
        num_few_shot: 25
    metrics:
    - type: acc_norm
      value: 25.17
      name: normalized accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: HellaSwag (10-Shot)
      type: hellaswag
      split: validation
      args:
        num_few_shot: 10
    metrics:
    - type: acc_norm
      value: 28.98
      name: normalized accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: MMLU (5-Shot)
      type: cais/mmlu
      config: all
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 26.17
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: TruthfulQA (0-shot)
      type: truthful_qa
      config: multiple_choice
      split: validation
      args:
        num_few_shot: 0
    metrics:
    - type: mc2
      value: 43.08
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: Winogrande (5-shot)
      type: winogrande
      config: winogrande_xl
      split: validation
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 52.01
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
  - task:
      type: text-generation
      name: Text Generation
    dataset:
      name: GSM8k (5-shot)
      type: gsm8k
      config: main
      split: test
      args:
        num_few_shot: 5
    metrics:
    - type: acc
      value: 0.61
      name: accuracy
    source:
      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=BEE-spoke-data/smol_llama-220M-openhermes
      name: Open LLM Leaderboard
---


# BEE-spoke-data/smol_llama-220M-openhermes

> Please note that this is an experiment, and the model has limitations because it is smol.


prompt format is alpaca


```
Below is an instruction that describes a task, paired with an input that
provides further context. Write a response that appropriately completes
the request.  

### Instruction:  

How can I increase my meme production/output? Currently, I only create them in ancient babylonian which is time consuming.  

### Inputs:

### Response:
```

It was trained on inputs so if you have inputs (like some text to ask a question about) then include it under `### Inputs:`


## Example

Output on the text above ^. The inference API is set to sample with low temp so you should see (_at least slightly_) different generations each time.


![image/png](https://cdn-uploads.huggingface.co/production/uploads/60bccec062080d33f875cd0c/0nFP2jsBkritnryKmI8NV.png)

Note that the inference API parameters used here are an initial educated guess, and may be updated over time:

```yml
inference:
  parameters:
    do_sample: true
    renormalize_logits: true
    temperature: 0.25
    top_p: 0.95
    top_k: 50
    min_new_tokens: 2
    max_new_tokens: 96
    repetition_penalty: 1.03
    no_repeat_ngram_size: 5
    epsilon_cutoff: 0.0008
```

Feel free to experiment with the parameters using the model in Python and let us know if you have improved results with other params!

## Data 

Note that **this checkpoint** was fine-tuned on `teknium/openhermes`, which is generated/synthetic data by an OpenAI model. This means usage of this checkpoint should follow their terms of use: https://openai.com/policies/terms-of-use  


---

# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_BEE-spoke-data__smol_llama-220M-openhermes)

|             Metric              |Value|
|---------------------------------|----:|
|Avg.                             |29.34|
|AI2 Reasoning Challenge (25-Shot)|25.17|
|HellaSwag (10-Shot)              |28.98|
|MMLU (5-Shot)                    |26.17|
|TruthfulQA (0-shot)              |43.08|
|Winogrande (5-shot)              |52.01|
|GSM8k (5-shot)                   | 0.61|