RichardErkhov
commited on
Commit
•
bb06af6
1
Parent(s):
1e3beae
uploaded readme
Browse files
README.md
ADDED
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Quantization made by Richard Erkhov.
|
2 |
+
|
3 |
+
[Github](https://github.com/RichardErkhov)
|
4 |
+
|
5 |
+
[Discord](https://discord.gg/pvy7H8DZMG)
|
6 |
+
|
7 |
+
[Request more models](https://github.com/RichardErkhov/quant_request)
|
8 |
+
|
9 |
+
|
10 |
+
Grafted-Hermetic-Platypus-C-2x7B - bnb 4bits
|
11 |
+
- Model creator: https://huggingface.co/lodrick-the-lafted/
|
12 |
+
- Original model: https://huggingface.co/lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B/
|
13 |
+
|
14 |
+
|
15 |
+
|
16 |
+
|
17 |
+
Original model description:
|
18 |
+
---
|
19 |
+
license: apache-2.0
|
20 |
+
datasets:
|
21 |
+
- lodrick-the-lafted/Hermes-217K
|
22 |
+
- garage-bAInd/Open-Platypus
|
23 |
+
- jondurbin/airoboros-3.2
|
24 |
+
model-index:
|
25 |
+
- name: Grafted-Hermetic-Platypus-C-2x7B
|
26 |
+
results:
|
27 |
+
- task:
|
28 |
+
type: text-generation
|
29 |
+
name: Text Generation
|
30 |
+
dataset:
|
31 |
+
name: AI2 Reasoning Challenge (25-Shot)
|
32 |
+
type: ai2_arc
|
33 |
+
config: ARC-Challenge
|
34 |
+
split: test
|
35 |
+
args:
|
36 |
+
num_few_shot: 25
|
37 |
+
metrics:
|
38 |
+
- type: acc_norm
|
39 |
+
value: 58.96
|
40 |
+
name: normalized accuracy
|
41 |
+
source:
|
42 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
43 |
+
name: Open LLM Leaderboard
|
44 |
+
- task:
|
45 |
+
type: text-generation
|
46 |
+
name: Text Generation
|
47 |
+
dataset:
|
48 |
+
name: HellaSwag (10-Shot)
|
49 |
+
type: hellaswag
|
50 |
+
split: validation
|
51 |
+
args:
|
52 |
+
num_few_shot: 10
|
53 |
+
metrics:
|
54 |
+
- type: acc_norm
|
55 |
+
value: 82.77
|
56 |
+
name: normalized accuracy
|
57 |
+
source:
|
58 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
59 |
+
name: Open LLM Leaderboard
|
60 |
+
- task:
|
61 |
+
type: text-generation
|
62 |
+
name: Text Generation
|
63 |
+
dataset:
|
64 |
+
name: MMLU (5-Shot)
|
65 |
+
type: cais/mmlu
|
66 |
+
config: all
|
67 |
+
split: test
|
68 |
+
args:
|
69 |
+
num_few_shot: 5
|
70 |
+
metrics:
|
71 |
+
- type: acc
|
72 |
+
value: 62.08
|
73 |
+
name: accuracy
|
74 |
+
source:
|
75 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
76 |
+
name: Open LLM Leaderboard
|
77 |
+
- task:
|
78 |
+
type: text-generation
|
79 |
+
name: Text Generation
|
80 |
+
dataset:
|
81 |
+
name: TruthfulQA (0-shot)
|
82 |
+
type: truthful_qa
|
83 |
+
config: multiple_choice
|
84 |
+
split: validation
|
85 |
+
args:
|
86 |
+
num_few_shot: 0
|
87 |
+
metrics:
|
88 |
+
- type: mc2
|
89 |
+
value: 60.87
|
90 |
+
source:
|
91 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
92 |
+
name: Open LLM Leaderboard
|
93 |
+
- task:
|
94 |
+
type: text-generation
|
95 |
+
name: Text Generation
|
96 |
+
dataset:
|
97 |
+
name: Winogrande (5-shot)
|
98 |
+
type: winogrande
|
99 |
+
config: winogrande_xl
|
100 |
+
split: validation
|
101 |
+
args:
|
102 |
+
num_few_shot: 5
|
103 |
+
metrics:
|
104 |
+
- type: acc
|
105 |
+
value: 77.74
|
106 |
+
name: accuracy
|
107 |
+
source:
|
108 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
109 |
+
name: Open LLM Leaderboard
|
110 |
+
- task:
|
111 |
+
type: text-generation
|
112 |
+
name: Text Generation
|
113 |
+
dataset:
|
114 |
+
name: GSM8k (5-shot)
|
115 |
+
type: gsm8k
|
116 |
+
config: main
|
117 |
+
split: test
|
118 |
+
args:
|
119 |
+
num_few_shot: 5
|
120 |
+
metrics:
|
121 |
+
- type: acc
|
122 |
+
value: 43.9
|
123 |
+
name: accuracy
|
124 |
+
source:
|
125 |
+
url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B
|
126 |
+
name: Open LLM Leaderboard
|
127 |
+
---
|
128 |
+
|
129 |
+
<img src=https://huggingface.co/lodrick-the-lafted/Grafted-Hermetic-Platypus-C-2x7B/resolve/main/ghp.png>
|
130 |
+
|
131 |
+
# Grafted-Hermetic-Platypus-C-2x7B
|
132 |
+
|
133 |
+
MoE merge of
|
134 |
+
- [Platyboros-Instruct-7B](https://huggingface.co/lodrick-the-lafted/Platyboros-Instruct-7B)
|
135 |
+
- [Hermes-Instruct-7B-217K](https://huggingface.co/lodrick-the-lafted/Hermes-Instruct-7B-217K)
|
136 |
+
|
137 |
+
<br />
|
138 |
+
<br />
|
139 |
+
|
140 |
+
# Prompt Format
|
141 |
+
|
142 |
+
Both the default Mistral-Instruct tags and Alpaca are fine, so either:
|
143 |
+
```
|
144 |
+
<s>[INST] {sys_prompt} {instruction} [/INST]
|
145 |
+
```
|
146 |
+
|
147 |
+
or
|
148 |
+
|
149 |
+
|
150 |
+
```
|
151 |
+
{sys_prompt}
|
152 |
+
|
153 |
+
### Instruction:
|
154 |
+
{instruction}
|
155 |
+
|
156 |
+
### Response:
|
157 |
+
|
158 |
+
```
|
159 |
+
The tokenizer default is Alpaca this time around.
|
160 |
+
|
161 |
+
<br />
|
162 |
+
<br />
|
163 |
+
|
164 |
+
# Usage
|
165 |
+
|
166 |
+
```python
|
167 |
+
from transformers import AutoTokenizer
|
168 |
+
import transformers
|
169 |
+
import torch
|
170 |
+
|
171 |
+
model = "lodrick-the-lafted/Grafted-Hermetic-Platypus-A-2x7B"
|
172 |
+
|
173 |
+
tokenizer = AutoTokenizer.from_pretrained(model)
|
174 |
+
pipeline = transformers.pipeline(
|
175 |
+
"text-generation",
|
176 |
+
model=model,
|
177 |
+
model_kwargs={"torch_dtype": torch.bfloat16},
|
178 |
+
)
|
179 |
+
|
180 |
+
messages = [{"role": "user", "content": "Give me a cooking recipe for an peach pie."}]
|
181 |
+
prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
182 |
+
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_p=0.95)
|
183 |
+
print(outputs[0]["generated_text"])
|
184 |
+
```
|
185 |
+
|
186 |
+
|
187 |
+
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
188 |
+
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_lodrick-the-lafted__Grafted-Hermetic-Platypus-C-2x7B)
|
189 |
+
|
190 |
+
| Metric |Value|
|
191 |
+
|---------------------------------|----:|
|
192 |
+
|Avg. |64.39|
|
193 |
+
|AI2 Reasoning Challenge (25-Shot)|58.96|
|
194 |
+
|HellaSwag (10-Shot) |82.77|
|
195 |
+
|MMLU (5-Shot) |62.08|
|
196 |
+
|TruthfulQA (0-shot) |60.87|
|
197 |
+
|Winogrande (5-shot) |77.74|
|
198 |
+
|GSM8k (5-shot) |43.90|
|
199 |
+
|
200 |
+
|
201 |
+
|