Text Generation
GGUF
English
fireplace
fireplace-2
valiant
valiant-labs
llama
llama-3.1
llama-3.1-instruct
llama-3.1-instruct-8b
llama-3
llama-3-instruct
llama-3-instruct-8b
8b
function-calling
sql
database
data-visualization
matplotlib
json
conversational
chat
instruct
Eval Results
Inference Endpoints
aashish1904
commited on
Commit
•
472589e
1
Parent(s):
7f102e8
Upload README.md with huggingface_hub
Browse files
README.md
ADDED
@@ -0,0 +1,262 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
---
|
3 |
+
|
4 |
+
language:
|
5 |
+
- en
|
6 |
+
license: llama3.1
|
7 |
+
tags:
|
8 |
+
- fireplace
|
9 |
+
- fireplace-2
|
10 |
+
- valiant
|
11 |
+
- valiant-labs
|
12 |
+
- llama
|
13 |
+
- llama-3.1
|
14 |
+
- llama-3.1-instruct
|
15 |
+
- llama-3.1-instruct-8b
|
16 |
+
- llama-3
|
17 |
+
- llama-3-instruct
|
18 |
+
- llama-3-instruct-8b
|
19 |
+
- 8b
|
20 |
+
- function-calling
|
21 |
+
- sql
|
22 |
+
- database
|
23 |
+
- data-visualization
|
24 |
+
- matplotlib
|
25 |
+
- json
|
26 |
+
- conversational
|
27 |
+
- chat
|
28 |
+
- instruct
|
29 |
+
pipeline_tag: text-generation
|
30 |
+
model_type: llama
|
31 |
+
model-index:
|
32 |
+
- name: Llama3.1-8B-Fireplace2
|
33 |
+
results:
|
34 |
+
- task:
|
35 |
+
type: text-generation
|
36 |
+
name: Text Generation
|
37 |
+
dataset:
|
38 |
+
name: IFEval (0-Shot)
|
39 |
+
type: HuggingFaceH4/ifeval
|
40 |
+
args:
|
41 |
+
num_few_shot: 0
|
42 |
+
metrics:
|
43 |
+
- type: inst_level_strict_acc and prompt_level_strict_acc
|
44 |
+
value: 54.83
|
45 |
+
name: strict accuracy
|
46 |
+
source:
|
47 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
48 |
+
name: Open LLM Leaderboard
|
49 |
+
- task:
|
50 |
+
type: text-generation
|
51 |
+
name: Text Generation
|
52 |
+
dataset:
|
53 |
+
name: BBH (3-Shot)
|
54 |
+
type: BBH
|
55 |
+
args:
|
56 |
+
num_few_shot: 3
|
57 |
+
metrics:
|
58 |
+
- type: acc_norm
|
59 |
+
value: 24.07
|
60 |
+
name: normalized accuracy
|
61 |
+
source:
|
62 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
63 |
+
name: Open LLM Leaderboard
|
64 |
+
- task:
|
65 |
+
type: text-generation
|
66 |
+
name: Text Generation
|
67 |
+
dataset:
|
68 |
+
name: MATH Lvl 5 (4-Shot)
|
69 |
+
type: hendrycks/competition_math
|
70 |
+
args:
|
71 |
+
num_few_shot: 4
|
72 |
+
metrics:
|
73 |
+
- type: exact_match
|
74 |
+
value: 5.82
|
75 |
+
name: exact match
|
76 |
+
source:
|
77 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
78 |
+
name: Open LLM Leaderboard
|
79 |
+
- task:
|
80 |
+
type: text-generation
|
81 |
+
name: Text Generation
|
82 |
+
dataset:
|
83 |
+
name: GPQA (0-shot)
|
84 |
+
type: Idavidrein/gpqa
|
85 |
+
args:
|
86 |
+
num_few_shot: 0
|
87 |
+
metrics:
|
88 |
+
- type: acc_norm
|
89 |
+
value: 5.15
|
90 |
+
name: acc_norm
|
91 |
+
source:
|
92 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
93 |
+
name: Open LLM Leaderboard
|
94 |
+
- task:
|
95 |
+
type: text-generation
|
96 |
+
name: Text Generation
|
97 |
+
dataset:
|
98 |
+
name: MuSR (0-shot)
|
99 |
+
type: TAUR-Lab/MuSR
|
100 |
+
args:
|
101 |
+
num_few_shot: 0
|
102 |
+
metrics:
|
103 |
+
- type: acc_norm
|
104 |
+
value: 4.38
|
105 |
+
name: acc_norm
|
106 |
+
source:
|
107 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
108 |
+
name: Open LLM Leaderboard
|
109 |
+
- task:
|
110 |
+
type: text-generation
|
111 |
+
name: Text Generation
|
112 |
+
dataset:
|
113 |
+
name: MMLU-PRO (5-shot)
|
114 |
+
type: TIGER-Lab/MMLU-Pro
|
115 |
+
config: main
|
116 |
+
split: test
|
117 |
+
args:
|
118 |
+
num_few_shot: 5
|
119 |
+
metrics:
|
120 |
+
- type: acc
|
121 |
+
value: 15.63
|
122 |
+
name: accuracy
|
123 |
+
source:
|
124 |
+
url: https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=ValiantLabs/Llama3.1-8B-Fireplace2
|
125 |
+
name: Open LLM Leaderboard
|
126 |
+
|
127 |
+
---
|
128 |
+
|
129 |
+
![](https://lh7-rt.googleusercontent.com/docsz/AD_4nXeiuCm7c8lEwEJuRey9kiVZsRn2W-b4pWlu3-X534V3YmVuVc2ZL-NXg2RkzSOOS2JXGHutDuyyNAUtdJI65jGTo8jT9Y99tMi4H4MqL44Uc5QKG77B0d6-JfIkZHFaUA71-RtjyYZWVIhqsNZcx8-OMaA?key=xt3VSDoCbmTY7o-cwwOFwQ)
|
130 |
+
|
131 |
+
# QuantFactory/Llama3.1-8B-Fireplace2-GGUF
|
132 |
+
This is quantized version of [ValiantLabs/Llama3.1-8B-Fireplace2](https://huggingface.co/ValiantLabs/Llama3.1-8B-Fireplace2) created using llama.cpp
|
133 |
+
|
134 |
+
# Original Model Card
|
135 |
+
|
136 |
+
|
137 |
+
|
138 |
+
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/64f267a8a4f79a118e0fcc89/JYkaXrk2DqpXhaL9WymKY.jpeg)
|
139 |
+
|
140 |
+
|
141 |
+
Fireplace 2 is a chat model, adding helpful structured outputs to Llama 3.1 8b Instruct.
|
142 |
+
- an expansion pack of supplementary outputs - request them at will within your chat:
|
143 |
+
- Inline function calls
|
144 |
+
- SQL queries
|
145 |
+
- JSON objects
|
146 |
+
- Data visualization with matplotlib
|
147 |
+
- Mix normal chat and structured outputs within the same conversation.
|
148 |
+
- Fireplace 2 supplements the existing strengths of Llama 3.1, providing inline capabilities within the Llama 3 Instruct format.
|
149 |
+
|
150 |
+
|
151 |
+
## Version
|
152 |
+
|
153 |
+
This is the **2024-07-23** release of Fireplace 2 for Llama 3.1 8b.
|
154 |
+
|
155 |
+
We're excited to bring further upgrades and releases to Fireplace 2 in the future.
|
156 |
+
|
157 |
+
Help us and recommend Fireplace 2 to your friends!
|
158 |
+
|
159 |
+
|
160 |
+
## Prompting Guide
|
161 |
+
Fireplace uses the [Llama 3.1 Instruct](https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct) prompt format. The example script below can be used as a starting point for general chat with Llama 3.1 and also includes the different special tokens used for Fireplace 2's added features:
|
162 |
+
|
163 |
+
|
164 |
+
import transformers
|
165 |
+
import torch
|
166 |
+
|
167 |
+
model_id = "ValiantLabs/Llama3.1-8B-Fireplace2"
|
168 |
+
|
169 |
+
pipeline = transformers.pipeline(
|
170 |
+
"text-generation",
|
171 |
+
model=model_id,
|
172 |
+
model_kwargs={"torch_dtype": torch.bfloat16},
|
173 |
+
device_map="auto",
|
174 |
+
)
|
175 |
+
|
176 |
+
messages = [
|
177 |
+
{"role": "system", "content": "You are Fireplace, an expert technical assistant."},
|
178 |
+
{"role": "user", "content": "Hi, can you explain local area networking to me?"}, #general Llama 3.1 chat
|
179 |
+
#{"role": "user", "content": "I have the following SQL table: employees (job_id VARCHAR, salary INTEGER)\n\nCan you find all employees with a salary above $75000?<|request_sql|>"}, #for SQL query
|
180 |
+
#{"role": "user", "content": "{""name"": ""get_news_headlines"",""description"": ""Get the latest news headlines"",""parameters"": {""type"": ""object"",""properties"": {""country"": {""type"": ""string"",""description"": ""The country for which news headlines are to be retrieved""}},""required"": [""country""]}}\n\nHi, can you get me the latest news headlines for the United States?<|request_function_call|>"}, # for function call
|
181 |
+
#{"role": "user", "content": "Show me an example of a histogram with a fixed bin size. Use attractive colors.<|request_matplotlib|>"}, #for data visualization
|
182 |
+
#{"role": "user", "content": "Can you define the word 'presence' for me, thanks!<|request_json|>"}, #for JSON output
|
183 |
+
]
|
184 |
+
|
185 |
+
outputs = pipeline(
|
186 |
+
messages,
|
187 |
+
max_new_tokens=512,
|
188 |
+
)
|
189 |
+
print(outputs[0]["generated_text"][-1])
|
190 |
+
|
191 |
+
|
192 |
+
While Fireplace 2 is trained to minimize incorrect structured outputs, they can still occur occasionally. Production uses of Fireplace 2 should verify the structure of all model outputs and remove any unneeded components of the output.
|
193 |
+
|
194 |
+
For handling of function call responses, use the [Llama 3.1 Instruct tool response style.](https://huggingface.co/blog/llama31#custom-tool-calling)
|
195 |
+
|
196 |
+
|
197 |
+
## Special Tokens
|
198 |
+
|
199 |
+
Fireplace 2 utilizes special tokens applied to the Llama 3.1 tokenizer:
|
200 |
+
|
201 |
+
- <|request_json|>
|
202 |
+
- <|start_json|>
|
203 |
+
- <|end_json|>
|
204 |
+
- <|request_sql|>
|
205 |
+
- <|start_sql|>
|
206 |
+
- <|end_sql|>
|
207 |
+
- <|request_matplotlib|>
|
208 |
+
- <|start_matplotlib|>
|
209 |
+
- <|end_matplotlib|>
|
210 |
+
- <|request_function_call|>
|
211 |
+
- <|start_function_call|>
|
212 |
+
- <|end_function_call|>
|
213 |
+
|
214 |
+
These are supplemental to the existing special tokens used by Llama 3.1, such as <|python_tag|> and <|start_header_id|>. Fireplace 2 has been trained using the Llama 3.1 Instruct chat structure, with new special tokens added within the conversation.
|
215 |
+
|
216 |
+
The 'request' tokens are used by the user to request a specific type of structured output. They should be appended to the end of the user's message and can be alternated with normal chat responses throughout the conversation.
|
217 |
+
|
218 |
+
|
219 |
+
## The Model
|
220 |
+
Fireplace 2 is built on top of Llama 3.1 8b Instruct.
|
221 |
+
|
222 |
+
This version of Fireplace 2 uses data from the following datasets:
|
223 |
+
|
224 |
+
- [glaiveai/glaive-function-calling-v2](https://huggingface.co/datasets/glaiveai/glaive-function-calling-v2)
|
225 |
+
- [b-mc2/sql-create-context](https://huggingface.co/datasets/b-mc2/sql-create-context)
|
226 |
+
- [sequelbox/Cadmium](https://huggingface.co/datasets/sequelbox/Cadmium)
|
227 |
+
- [sequelbox/Harlequin](https://huggingface.co/datasets/sequelbox/Harlequin)
|
228 |
+
- [migtissera/Tess-v1.5](https://huggingface.co/datasets/migtissera/Tess-v1.5)
|
229 |
+
- [LDJnr/Pure-Dove](https://huggingface.co/datasets/LDJnr/Pure-Dove)
|
230 |
+
|
231 |
+
Additional capabilities will be added to future releases.
|
232 |
+
|
233 |
+
|
234 |
+
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
|
235 |
+
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_ValiantLabs__Llama3.1-8B-Fireplace2)
|
236 |
+
|
237 |
+
| Metric |Value|
|
238 |
+
|-------------------|----:|
|
239 |
+
|Avg. |18.31|
|
240 |
+
|IFEval (0-Shot) |54.83|
|
241 |
+
|BBH (3-Shot) |24.07|
|
242 |
+
|MATH Lvl 5 (4-Shot)| 5.82|
|
243 |
+
|GPQA (0-shot) | 5.15|
|
244 |
+
|MuSR (0-shot) | 4.38|
|
245 |
+
|MMLU-PRO (5-shot) |15.63|
|
246 |
+
|
247 |
+
|
248 |
+
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/63444f2687964b331809eb55/VCJ8Fmefd8cdVhXSSxJiD.jpeg)
|
249 |
+
|
250 |
+
|
251 |
+
Fireplace 2 is created by [Valiant Labs.](http://valiantlabs.ca/)
|
252 |
+
|
253 |
+
[Check out our HuggingFace page for Shining Valiant 2 and our other models!](https://huggingface.co/ValiantLabs)
|
254 |
+
|
255 |
+
[Follow us on X for updates on our models!](https://twitter.com/valiant_labs)
|
256 |
+
|
257 |
+
We care about open source.
|
258 |
+
For everyone to use.
|
259 |
+
|
260 |
+
We encourage others to finetune further from our models.
|
261 |
+
|
262 |
+
|