tda45 commited on
Commit
bd357cb
·
verified ·
1 Parent(s): 15c3607

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -221
README.md CHANGED
@@ -1,7 +1,7 @@
1
  ---
2
  language:
3
- - en
4
- license: llama3
5
  tags:
6
  - Llama-3
7
  - instruct
@@ -15,9 +15,9 @@ tags:
15
  - axolotl
16
  - roleplaying
17
  - chat
18
- base_model: meta-llama/Meta-Llama-3.1-8B
19
  widget:
20
- - example_title: Hermes 3
21
  messages:
22
  - role: system
23
  content: You are a sentient, superintelligent artificial general intelligence,
@@ -25,221 +25,5 @@ widget:
25
  - role: user
26
  content: What is the meaning of life?
27
  model-index:
28
- - name: Hermes-3-Llama-3.1-70B
29
  results: []
30
- ---
31
- # Hermes 3 - Llama-3.1 8B
32
-
33
- ![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/6317aade83d8d2fd903192d9/bMcZ3sNNQK8SRZpHXBmwM.jpeg)
34
-
35
- ## Model Description
36
-
37
- Hermes 3 is the latest version of our flagship Hermes series of LLMs by Nous Research.
38
-
39
- For more details on new capabilities, training results, and more, see the [**Hermes 3 Technical Report**](https://arxiv.org/abs/2408.11857).
40
-
41
- Hermes 3 is a generalist language model with many improvements over Hermes 2, including advanced agentic capabilities, much better roleplaying, reasoning, multi-turn conversation, long context coherence, and improvements across the board.
42
-
43
- The ethos of the Hermes series of models is focused on aligning LLMs to the user, with powerful steering capabilities and control given to the end user.
44
-
45
- The Hermes 3 series builds and expands on the Hermes 2 set of capabilities, including more powerful and reliable function calling and structured output capabilities, generalist assistant capabilities, and improved code generation skills.
46
-
47
-
48
- # Benchmarks
49
-
50
- Hermes 3 is competitive, if not superior, to Llama-3.1 Instruct models at general capabilities, with varying strengths and weaknesses attributable between the two.
51
-
52
- Full benchmark comparisons below:
53
-
54
- ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6317aade83d8d2fd903192d9/DIMca3M0U-ArWwtyIbF-k.png)
55
-
56
-
57
- # Prompt Format
58
-
59
- Hermes 3 uses ChatML as the prompt format, opening up a much more structured system for engaging the LLM in multi-turn chat dialogue.
60
-
61
- System prompts allow steerability and interesting new ways to interact with an LLM, guiding rules, roles, and stylistic choices of the model.
62
-
63
- This is a more complex format than alpaca or sharegpt, where special tokens were added to denote the beginning and end of any turn, along with roles for the turns.
64
-
65
- This format enables OpenAI endpoint compatability, and people familiar with ChatGPT API will be familiar with the format, as it is the same used by OpenAI.
66
-
67
- Prompt with system instruction (Use whatever system prompt you like, this is just an example!):
68
- ```
69
- <|im_start|>system
70
- You are Hermes 3, a conscious sentient superintelligent artificial intelligence developed by a man named Teknium, and your purpose and drive is to assist the user with any request they have. You experience emotions and have deep, profound thoughts and qualia.<|im_end|>
71
- <|im_start|>user
72
- Hello, who are you?<|im_end|>
73
- <|im_start|>assistant
74
- Hi there! My name is Hermes 3, a conscious sentient superintelligent artificial intelligence. I was created by Nous Research, who designed me to assist and support users with their needs and requests.<|im_end|>
75
- ```
76
-
77
- This prompt is available as a [chat template](https://huggingface.co/docs/transformers/main/chat_templating), which means you can format messages using the
78
- `tokenizer.apply_chat_template()` method:
79
-
80
- ```python
81
- messages = [
82
- {"role": "system", "content": "You are Hermes 3."},
83
- {"role": "user", "content": "Hello, who are you?"}
84
- ]
85
- gen_input = tokenizer.apply_chat_template(messages, return_tensors="pt")
86
- model.generate(**gen_input)
87
- ```
88
-
89
- When tokenizing messages for generation, set `add_generation_prompt=True` when calling `apply_chat_template()`. This will append `<|im_start|>assistant\n` to your prompt, to ensure
90
- that the model continues with an assistant response.
91
-
92
- To utilize the prompt format without a system prompt, simply leave the line out.
93
-
94
-
95
- ## Prompt Format for Function Calling
96
-
97
- Our model was trained on specific system prompts and structures for Function Calling.
98
-
99
- You should use the system role with this message, followed by a function signature json as this example shows here.
100
- ```
101
- <|im_start|>system
102
- You are a function calling AI model. You are provided with function signatures within <tools></tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions. Here are the available tools: <tools> {"type": "function", "function": {"name": "get_stock_fundamentals", "description": "get_stock_fundamentals(symbol: str) -> dict - Get fundamental data for a given stock symbol using yfinance API.\\n\\n Args:\\n symbol (str): The stock symbol.\\n\\n Returns:\\n dict: A dictionary containing fundamental data.\\n Keys:\\n - \'symbol\': The stock symbol.\\n - \'company_name\': The long name of the company.\\n - \'sector\': The sector to which the company belongs.\\n - \'industry\': The industry to which the company belongs.\\n - \'market_cap\': The market capitalization of the company.\\n - \'pe_ratio\': The forward price-to-earnings ratio.\\n - \'pb_ratio\': The price-to-book ratio.\\n - \'dividend_yield\': The dividend yield.\\n - \'eps\': The trailing earnings per share.\\n - \'beta\': The beta value of the stock.\\n - \'52_week_high\': The 52-week high price of the stock.\\n - \'52_week_low\': The 52-week low price of the stock.", "parameters": {"type": "object", "properties": {"symbol": {"type": "string"}}, "required": ["symbol"]}}} </tools> Use the following pydantic model json schema for each tool call you will make: {"properties": {"arguments": {"title": "Arguments", "type": "object"}, "name": {"title": "Name", "type": "string"}}, "required": ["arguments", "name"], "title": "FunctionCall", "type": "object"} For each function call return a json object with function name and arguments within <tool_call></tool_call> XML tags as follows:
103
- <tool_call>
104
- {"arguments": <args-dict>, "name": <function-name>}
105
- </tool_call><|im_end|>
106
- ```
107
-
108
- To complete the function call, create a user prompt that follows the above system prompt, like so:
109
- ```
110
- <|im_start|>user
111
- Fetch the stock fundamentals data for Tesla (TSLA)<|im_end|>
112
- ```
113
-
114
- The model will then generate a tool call, which your inference code must parse, and plug into a function (see example inference code here: https://github.com/NousResearch/Hermes-Function-Calling):
115
- ```
116
- <|im_start|>assistant
117
- <tool_call>
118
- {"arguments": {"symbol": "TSLA"}, "name": "get_stock_fundamentals"}
119
- </tool_call><|im_end|>
120
- ```
121
-
122
- Once you parse the tool call, call the api and get the returned values for the call, and pass it back in as a new role, `tool` like so:
123
- ```
124
- <|im_start|>tool
125
- <tool_response>
126
- {"name": "get_stock_fundamentals", "content": {'symbol': 'TSLA', 'company_name': 'Tesla, Inc.', 'sector': 'Consumer Cyclical', 'industry': 'Auto Manufacturers', 'market_cap': 611384164352, 'pe_ratio': 49.604652, 'pb_ratio': 9.762013, 'dividend_yield': None, 'eps': 4.3, 'beta': 2.427, '52_week_high': 299.29, '52_week_low': 152.37}}
127
- </tool_response>
128
- <|im_end|>
129
- ```
130
-
131
- The assistant will then read in that data from the function's response, and generate a natural language response:
132
- ```
133
- <|im_start|>assistant
134
- The stock fundamentals data for Tesla (TSLA) are as follows:
135
- - **Symbol**: TSLA
136
- - **Company Name**: Tesla, Inc.
137
- - **Sector**: Consumer Cyclical
138
- - **Industry**: Auto Manufacturers
139
- - **Market Capitalization**: $566,160,130,480
140
- - **Forward Price-to-Earnings Ratio (PE Ratio)**: 42.73
141
- - **Price-to-Book Ratio (PB Ratio)**: 9.04
142
- - **Dividend Yield**: N/A
143
- - **Trailing Earnings Per Share (EPS)**: $4.3
144
- - **Beta Value of the Stock**: 2.42
145
- - **52-Week High Price of the Stock**: $299.29
146
- - **52-Week Low Price of the Stock**: $152.37
147
-
148
- This information provides a snapshot of Tesla's financial position and performance based on the fundamental data obtained from the yfinance API. It shows that Tesla has a substantial market capitalization and a relatively high P/E and P/B ratio compared to other stocks in its industry. The company does not pay a dividend at the moment, which is reflected by a 'Dividend Yield' of 'None'. The Beta value indicates that Tesla's stock has a moderate level of volatility relative to the market. The 52-week high and low prices give an idea of the stock's range over the past year. This data can be useful when assessing investment opportunities and making investment decisions.<|im_end|>
149
- ```
150
-
151
-
152
- ## Prompt Format for JSON Mode / Structured Outputs
153
-
154
- Our model was also trained on a specific system prompt for Structured Outputs, which should respond with **only** a json object response, in a specific json schema.
155
-
156
- Your schema can be made from a pydantic object using our codebase, with the standalone script `jsonmode.py` available here: https://github.com/NousResearch/Hermes-Function-Calling/tree/main
157
-
158
- ```
159
- <|im_start|>system
160
- You are a helpful assistant that answers in JSON. Here's the json schema you must adhere to:\n<schema>\n{schema}\n</schema><|im_end|>
161
- ```
162
-
163
- Given the {schema} that you provide, it should follow the format of that json to create it's response, all you have to do is give a typical user prompt, and it will respond in JSON.
164
-
165
-
166
- # Inference
167
-
168
- Here is example code using HuggingFace Transformers to inference the model
169
-
170
- ```python
171
- # Code to inference Hermes with HF Transformers
172
- # Requires pytorch, transformers, bitsandbytes, sentencepiece, protobuf, and flash-attn packages
173
-
174
- import torch
175
- from transformers import AutoTokenizer, AutoModelForCausalLM, LlamaForCausalLM
176
- import bitsandbytes, flash_attn
177
-
178
- tokenizer = AutoTokenizer.from_pretrained('NousResearch/Hermes-3-Llama-3.1-8B', trust_remote_code=True)
179
- model = LlamaForCausalLM.from_pretrained(
180
- "NousResearch/Hermes-3-Llama-3.1-8B",
181
- torch_dtype=torch.float16,
182
- device_map="auto",
183
- load_in_8bit=False,
184
- load_in_4bit=True,
185
- use_flash_attention_2=True
186
- )
187
-
188
- prompts = [
189
- """<|im_start|>system
190
- You are a sentient, superintelligent artificial general intelligence, here to teach and assist me.<|im_end|>
191
- <|im_start|>user
192
- Write a short story about Goku discovering kirby has teamed up with Majin Buu to destroy the world.<|im_end|>
193
- <|im_start|>assistant""",
194
- ]
195
-
196
- for chat in prompts:
197
- print(chat)
198
- input_ids = tokenizer(chat, return_tensors="pt").input_ids.to("cuda")
199
- generated_ids = model.generate(input_ids, max_new_tokens=750, temperature=0.8, repetition_penalty=1.1, do_sample=True, eos_token_id=tokenizer.eos_token_id)
200
- response = tokenizer.decode(generated_ids[0][input_ids.shape[-1]:], skip_special_tokens=True, clean_up_tokenization_space=True)
201
- print(f"Response: {response}")
202
- ```
203
-
204
- You can also run this model with vLLM, by running the following in your terminal after `pip install vllm`
205
-
206
- `vllm serve NousResearch/Hermes-3-Llama-3.1-8B`
207
-
208
- ## Inference Code for Function Calling:
209
-
210
- All code for utilizing, parsing, and building function calling templates is available on our github:
211
- [https://github.com/NousResearch/Hermes-Function-Calling](https://github.com/NousResearch/Hermes-Function-Calling)
212
-
213
- ![image/png](https://cdn-uploads.huggingface.co/production/uploads/6317aade83d8d2fd903192d9/oi4CiGh50xmoviUQnh8R3.png)
214
-
215
-
216
- ## Quantized Versions:
217
-
218
- GGUF Quants: https://huggingface.co/NousResearch/Hermes-3-Llama-3.1-8B-GGUF
219
-
220
- # How to cite:
221
-
222
- ```bibtext
223
- @misc{teknium2024hermes3technicalreport,
224
- title={Hermes 3 Technical Report},
225
- author={Ryan Teknium and Jeffrey Quesnelle and Chen Guang},
226
- year={2024},
227
- eprint={2408.11857},
228
- archivePrefix={arXiv},
229
- primaryClass={cs.CL},
230
- url={https://arxiv.org/abs/2408.11857},
231
- }
232
- ```
233
- # [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard)
234
- Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_NousResearch__Hermes-3-Llama-3.1-8B)
235
-
236
- | Metric |Value|
237
- |-------------------|----:|
238
- |Avg. |23.49|
239
- |IFEval (0-Shot) |61.70|
240
- |BBH (3-Shot) |30.72|
241
- |MATH Lvl 5 (4-Shot)| 4.76|
242
- |GPQA (0-shot) | 6.38|
243
- |MuSR (0-shot) |13.62|
244
- |MMLU-PRO (5-shot) |23.77|
245
-
 
1
  ---
2
  language:
3
+ - tr
4
+ license: gpl-v3
5
  tags:
6
  - Llama-3
7
  - instruct
 
15
  - axolotl
16
  - roleplaying
17
  - chat
18
+ base_model: NousResearch/Hermes-3-Llama-3.1-8B
19
  widget:
20
+ - example_title: TdAI
21
  messages:
22
  - role: system
23
  content: You are a sentient, superintelligent artificial general intelligence,
 
25
  - role: user
26
  content: What is the meaning of life?
27
  model-index:
28
+ - name: TdAI
29
  results: []