TheBloke commited on
Commit
7db5b6c
1 Parent(s): f5ba65e

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +381 -0
README.md ADDED
@@ -0,0 +1,381 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: NurtureAI/SynthIA-7B-v2.0-16k
3
+ inference: false
4
+ language:
5
+ - en
6
+ library_name: transformers
7
+ license: apache-2.0
8
+ model_creator: NurtureAI
9
+ model_name: SynthIA 7B V2.0 16K
10
+ model_type: mistral
11
+ pipeline_tag: text-generation
12
+ prompt_template: 'SYSTEM: Elaborate on the topic using a Tree of Thoughts and backtrack
13
+ when necessary to construct a clear, cohesive Chain of Thought reasoning. Always
14
+ answer without hesitation.
15
+
16
+ USER: {prompt}
17
+
18
+ ASSISTANT:
19
+
20
+ '
21
+ quantized_by: TheBloke
22
+ ---
23
+ <!-- markdownlint-disable MD041 -->
24
+
25
+ <!-- header start -->
26
+ <!-- 200823 -->
27
+ <div style="width: auto; margin-left: auto; margin-right: auto">
28
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
29
+ </div>
30
+ <div style="display: flex; justify-content: space-between; width: 100%;">
31
+ <div style="display: flex; flex-direction: column; align-items: flex-start;">
32
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
33
+ </div>
34
+ <div style="display: flex; flex-direction: column; align-items: flex-end;">
35
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
36
+ </div>
37
+ </div>
38
+ <div style="text-align:center; margin-top: 0em; margin-bottom: 0em"><p style="margin-top: 0.25em; margin-bottom: 0em;">TheBloke's LLM work is generously supported by a grant from <a href="https://a16z.com">andreessen horowitz (a16z)</a></p></div>
39
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
40
+ <!-- header end -->
41
+
42
+ # SynthIA 7B V2.0 16K - GGUF
43
+ - Model creator: [NurtureAI](https://huggingface.co/NurtureAI)
44
+ - Original model: [SynthIA 7B V2.0 16K](https://huggingface.co/NurtureAI/SynthIA-7B-v2.0-16k)
45
+
46
+ <!-- description start -->
47
+ ## Description
48
+
49
+ This repo contains GGUF format model files for [NurtureAI's SynthIA 7B V2.0 16K](https://huggingface.co/NurtureAI/SynthIA-7B-v2.0-16k).
50
+
51
+ These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
52
+
53
+ <!-- description end -->
54
+ <!-- README_GGUF.md-about-gguf start -->
55
+ ### About GGUF
56
+
57
+ GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
58
+
59
+ Here is an incomplete list of clients and libraries that are known to support GGUF:
60
+
61
+ * [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
62
+ * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
63
+ * [KoboldCpp](https://github.com/LostRuins/koboldcpp), a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
64
+ * [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
65
+ * [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), a great web UI with many interesting and unique features, including a full model library for easy model selection.
66
+ * [Faraday.dev](https://faraday.dev/), an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
67
+ * [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
68
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
69
+ * [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
70
+
71
+ <!-- README_GGUF.md-about-gguf end -->
72
+ <!-- repositories-available start -->
73
+ ## Repositories available
74
+
75
+ * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-AWQ)
76
+ * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GPTQ)
77
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF)
78
+ * [NurtureAI's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/NurtureAI/SynthIA-7B-v2.0-16k)
79
+ <!-- repositories-available end -->
80
+
81
+ <!-- prompt-template start -->
82
+ ## Prompt template: Synthia-CoT
83
+
84
+ ```
85
+ SYSTEM: Elaborate on the topic using a Tree of Thoughts and backtrack when necessary to construct a clear, cohesive Chain of Thought reasoning. Always answer without hesitation.
86
+ USER: {prompt}
87
+ ASSISTANT:
88
+
89
+ ```
90
+
91
+ <!-- prompt-template end -->
92
+
93
+
94
+ <!-- compatibility_gguf start -->
95
+ ## Compatibility
96
+
97
+ These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221)
98
+
99
+ They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
100
+
101
+ ## Explanation of quantisation methods
102
+
103
+ <details>
104
+ <summary>Click to see details</summary>
105
+
106
+ The new methods available are:
107
+
108
+ * GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
109
+ * GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
110
+ * GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
111
+ * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
112
+ * GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
113
+
114
+ Refer to the Provided Files table below to see what files use which methods, and how.
115
+ </details>
116
+ <!-- compatibility_gguf end -->
117
+
118
+ <!-- README_GGUF.md-provided-files start -->
119
+ ## Provided files
120
+
121
+ | Name | Quant method | Bits | Size | Max RAM required | Use case |
122
+ | ---- | ---- | ---- | ---- | ---- | ----- |
123
+ | [synthia-7b-v2.0-16k.Q2_K.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q2_K.gguf) | Q2_K | 2 | 3.08 GB| 5.58 GB | smallest, significant quality loss - not recommended for most purposes |
124
+ | [synthia-7b-v2.0-16k.Q3_K_S.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q3_K_S.gguf) | Q3_K_S | 3 | 3.16 GB| 5.66 GB | very small, high quality loss |
125
+ | [synthia-7b-v2.0-16k.Q3_K_M.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q3_K_M.gguf) | Q3_K_M | 3 | 3.52 GB| 6.02 GB | very small, high quality loss |
126
+ | [synthia-7b-v2.0-16k.Q3_K_L.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q3_K_L.gguf) | Q3_K_L | 3 | 3.82 GB| 6.32 GB | small, substantial quality loss |
127
+ | [synthia-7b-v2.0-16k.Q4_0.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q4_0.gguf) | Q4_0 | 4 | 4.11 GB| 6.61 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
128
+ | [synthia-7b-v2.0-16k.Q4_K_S.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q4_K_S.gguf) | Q4_K_S | 4 | 4.14 GB| 6.64 GB | small, greater quality loss |
129
+ | [synthia-7b-v2.0-16k.Q4_K_M.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q4_K_M.gguf) | Q4_K_M | 4 | 4.37 GB| 6.87 GB | medium, balanced quality - recommended |
130
+ | [synthia-7b-v2.0-16k.Q5_0.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q5_0.gguf) | Q5_0 | 5 | 5.00 GB| 7.50 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
131
+ | [synthia-7b-v2.0-16k.Q5_K_S.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q5_K_S.gguf) | Q5_K_S | 5 | 5.00 GB| 7.50 GB | large, low quality loss - recommended |
132
+ | [synthia-7b-v2.0-16k.Q5_K_M.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q5_K_M.gguf) | Q5_K_M | 5 | 5.13 GB| 7.63 GB | large, very low quality loss - recommended |
133
+ | [synthia-7b-v2.0-16k.Q6_K.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q6_K.gguf) | Q6_K | 6 | 5.94 GB| 8.44 GB | very large, extremely low quality loss |
134
+ | [synthia-7b-v2.0-16k.Q8_0.gguf](https://huggingface.co/TheBloke/SynthIA-7B-v2.0-16k-GGUF/blob/main/synthia-7b-v2.0-16k.Q8_0.gguf) | Q8_0 | 8 | 7.70 GB| 10.20 GB | very large, extremely low quality loss - not recommended |
135
+
136
+ **Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
137
+
138
+
139
+
140
+ <!-- README_GGUF.md-provided-files end -->
141
+
142
+ <!-- README_GGUF.md-how-to-download start -->
143
+ ## How to download GGUF files
144
+
145
+ **Note for manual downloaders:** You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
146
+
147
+ The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
148
+
149
+ * LM Studio
150
+ * LoLLMS Web UI
151
+ * Faraday.dev
152
+
153
+ ### In `text-generation-webui`
154
+
155
+ Under Download Model, you can enter the model repo: TheBloke/SynthIA-7B-v2.0-16k-GGUF and below it, a specific filename to download, such as: synthia-7b-v2.0-16k.Q4_K_M.gguf.
156
+
157
+ Then click Download.
158
+
159
+ ### On the command line, including multiple files at once
160
+
161
+ I recommend using the `huggingface-hub` Python library:
162
+
163
+ ```shell
164
+ pip3 install huggingface-hub
165
+ ```
166
+
167
+ Then you can download any individual model file to the current directory, at high speed, with a command like this:
168
+
169
+ ```shell
170
+ huggingface-cli download TheBloke/SynthIA-7B-v2.0-16k-GGUF synthia-7b-v2.0-16k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
171
+ ```
172
+
173
+ <details>
174
+ <summary>More advanced huggingface-cli download usage</summary>
175
+
176
+ You can also download multiple files at once with a pattern:
177
+
178
+ ```shell
179
+ huggingface-cli download TheBloke/SynthIA-7B-v2.0-16k-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
180
+ ```
181
+
182
+ For more documentation on downloading with `huggingface-cli`, please see: [HF -> Hub Python Library -> Download files -> Download from the CLI](https://huggingface.co/docs/huggingface_hub/guides/download#download-from-the-cli).
183
+
184
+ To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
185
+
186
+ ```shell
187
+ pip3 install hf_transfer
188
+ ```
189
+
190
+ And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
191
+
192
+ ```shell
193
+ HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/SynthIA-7B-v2.0-16k-GGUF synthia-7b-v2.0-16k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
194
+ ```
195
+
196
+ Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
197
+ </details>
198
+ <!-- README_GGUF.md-how-to-download end -->
199
+
200
+ <!-- README_GGUF.md-how-to-run start -->
201
+ ## Example `llama.cpp` command
202
+
203
+ Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
204
+
205
+ ```shell
206
+ ./main -ngl 32 -m synthia-7b-v2.0-16k.Q4_K_M.gguf --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "SYSTEM: Elaborate on the topic using a Tree of Thoughts and backtrack when necessary to construct a clear, cohesive Chain of Thought reasoning. Always answer without hesitation.\nUSER: {prompt}\nASSISTANT:"
207
+ ```
208
+
209
+ Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
210
+
211
+ Change `-c 2048` to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.
212
+
213
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
214
+
215
+ For other parameters and how to use them, please refer to [the llama.cpp documentation](https://github.com/ggerganov/llama.cpp/blob/master/examples/main/README.md)
216
+
217
+ ## How to run in `text-generation-webui`
218
+
219
+ Further instructions can be found in the text-generation-webui documentation, here: [text-generation-webui/docs/04 ‐ Model Tab.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/04%20%E2%80%90%20Model%20Tab.md#llamacpp).
220
+
221
+ ## How to run from Python code
222
+
223
+ You can use GGUF models from Python using the [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) or [ctransformers](https://github.com/marella/ctransformers) libraries.
224
+
225
+ ### How to load this model in Python code, using ctransformers
226
+
227
+ #### First install the package
228
+
229
+ Run one of the following commands, according to your system:
230
+
231
+ ```shell
232
+ # Base ctransformers with no GPU acceleration
233
+ pip install ctransformers
234
+ # Or with CUDA GPU acceleration
235
+ pip install ctransformers[cuda]
236
+ # Or with AMD ROCm GPU acceleration (Linux only)
237
+ CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
238
+ # Or with Metal GPU acceleration for macOS systems only
239
+ CT_METAL=1 pip install ctransformers --no-binary ctransformers
240
+ ```
241
+
242
+ #### Simple ctransformers example code
243
+
244
+ ```python
245
+ from ctransformers import AutoModelForCausalLM
246
+
247
+ # Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
248
+ llm = AutoModelForCausalLM.from_pretrained("TheBloke/SynthIA-7B-v2.0-16k-GGUF", model_file="synthia-7b-v2.0-16k.Q4_K_M.gguf", model_type="mistral", gpu_layers=50)
249
+
250
+ print(llm("AI is going to"))
251
+ ```
252
+
253
+ ## How to use with LangChain
254
+
255
+ Here are guides on using llama-cpp-python and ctransformers with LangChain:
256
+
257
+ * [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
258
+ * [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
259
+
260
+ <!-- README_GGUF.md-how-to-run end -->
261
+
262
+ <!-- footer start -->
263
+ <!-- 200823 -->
264
+ ## Discord
265
+
266
+ For further support, and discussions on these models and AI in general, join us at:
267
+
268
+ [TheBloke AI's Discord server](https://discord.gg/theblokeai)
269
+
270
+ ## Thanks, and how to contribute
271
+
272
+ Thanks to the [chirper.ai](https://chirper.ai) team!
273
+
274
+ Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
275
+
276
+ I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
277
+
278
+ If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
279
+
280
+ Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
281
+
282
+ * Patreon: https://patreon.com/TheBlokeAI
283
+ * Ko-Fi: https://ko-fi.com/TheBlokeAI
284
+
285
+ **Special thanks to**: Aemon Algiz.
286
+
287
+ **Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
288
+
289
+
290
+ Thank you to all my generous patrons and donaters!
291
+
292
+ And thank you again to a16z for their generous grant.
293
+
294
+ <!-- footer end -->
295
+
296
+ <!-- original-model-card start -->
297
+ # Original model card: NurtureAI's SynthIA 7B V2.0 16K
298
+
299
+
300
+ ## SynthIA 7B v2.0 extended to 16k context
301
+
302
+ This is the same original model so the original model license still applies.
303
+
304
+ This model has been modified to have a larger maximum context size of 16k.
305
+
306
+ ### Original Model Card:
307
+
308
+ ### Prompt format:
309
+
310
+ ```
311
+ SYSTEM: Elaborate on the topic using a Tree of Thoughts and backtrack when necessary to construct a clear, cohesive Chain of Thought reasoning. Always answer without hesitation.
312
+ USER: How is insulin synthesized?
313
+ ASSISTANT:
314
+ ```
315
+
316
+ ### Code example:
317
+
318
+ ```python
319
+ from transformers import AutoModelForCausalLM, AutoTokenizer
320
+ import torch, json
321
+ # model path
322
+ model_path = "NurtureAI/SynthIA-7B-v2.0-16k"
323
+ output_file_path = "./SynthIA-7B-v2.0-conversations.jsonl"
324
+ device_map = {"": "cuda"}
325
+ model = AutoModelForCausalLM.from_pretrained(
326
+ model_path,
327
+ torch_dtype=torch.float16,
328
+ device_map=device_map,
329
+ load_in_8bit=False,
330
+ trust_remote_code=True,
331
+ )
332
+ # tokenizer
333
+ tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
334
+
335
+ def generate_text(instruction):
336
+ tokens = tokenizer.encode(instruction)
337
+ tokens = torch.LongTensor(tokens).unsqueeze(0)
338
+ tokens = tokens.to("cuda")
339
+ instance = {
340
+ "input_ids": tokens,
341
+ "top_p": 1.0,
342
+ "temperature": 0.75,
343
+ "generate_len": 1024,
344
+ "top_k": 50,
345
+ }
346
+ length = len(tokens[0])
347
+ with torch.no_grad():
348
+ rest = model.generate(
349
+ input_ids=tokens,
350
+ max_length=length + instance["generate_len"],
351
+ use_cache=True,
352
+ do_sample=True,
353
+ top_p=instance["top_p"],
354
+ temperature=instance["temperature"],
355
+ top_k=instance["top_k"],
356
+ num_return_sequences=1,
357
+ )
358
+ output = rest[0][length:]
359
+ string = tokenizer.decode(output, skip_special_tokens=True)
360
+ answer = string.split("USER:")[0].strip()
361
+ return f"{answer}"
362
+
363
+
364
+ conversation = f"SYSTEM: Elaborate on the topic using a Tree of Thoughts and backtrack when necessary to construct a clear, cohesive Chain of Thought reasoning. Always answer without hesitation."
365
+
366
+
367
+ while True:
368
+ user_input = input("You: ")
369
+ llm_prompt = f"{conversation} \nUSER: {user_input} \nASSISTANT: "
370
+ answer = generate_text(llm_prompt)
371
+ print(answer)
372
+ conversation = f"{llm_prompt}{answer}"
373
+ json_data = {"prompt": user_input, "answer": answer}
374
+
375
+ ## Save your conversation
376
+ with open(output_file_path, "a") as output_file:
377
+ output_file.write(json.dumps(json_data) + "\n")
378
+
379
+ ```
380
+
381
+ <!-- original-model-card end -->