Text Generation
Transformers
GGUF
English
llama
TheBloke commited on
Commit
f6c1648
1 Parent(s): fbe11f4

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +409 -0
README.md ADDED
@@ -0,0 +1,409 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets:
3
+ - psmathur/orca_mini_v1_dataset
4
+ - ehartford/dolphin
5
+ inference: false
6
+ language:
7
+ - en
8
+ library_name: transformers
9
+ license: llama2
10
+ model_creator: Pankaj Mathur
11
+ model_link: https://huggingface.co/psmathur/orca_mini_v3_70b
12
+ model_name: Orca Mini v3 70B
13
+ model_type: llama
14
+ pipeline_tag: text-generation
15
+ quantized_by: TheBloke
16
+ ---
17
+
18
+ <!-- header start -->
19
+ <!-- 200823 -->
20
+ <div style="width: auto; margin-left: auto; margin-right: auto">
21
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
22
+ </div>
23
+ <div style="display: flex; justify-content: space-between; width: 100%;">
24
+ <div style="display: flex; flex-direction: column; align-items: flex-start;">
25
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
26
+ </div>
27
+ <div style="display: flex; flex-direction: column; align-items: flex-end;">
28
+ <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>
29
+ </div>
30
+ </div>
31
+ <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>
32
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
33
+ <!-- header end -->
34
+
35
+ # Orca Mini v3 70B - GGUF
36
+ - Model creator: [Pankaj Mathur](https://huggingface.co/psmathur)
37
+ - Original model: [Orca Mini v3 70B](https://huggingface.co/psmathur/orca_mini_v3_70b)
38
+
39
+ ## Description
40
+
41
+ This repo contains GGUF format model files for [Pankaj Mathur's Orca Mini v3 70B](https://huggingface.co/psmathur/orca_mini_v3_70b).
42
+
43
+ <!-- README_GGUF.md-about-gguf start -->
44
+ ### About GGUF
45
+
46
+ 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.
47
+
48
+ The key benefit of GGUF is that it is a extensible, future-proof format which stores more information about the model as metadata. It also includes significantly improved tokenization code, including for the first time full support for special tokens. This should improve performance, especially with models that use new special tokens and implement custom prompt templates.
49
+
50
+ Here are a list of clients and libraries that are known to support GGUF:
51
+ * [llama.cpp](https://github.com/ggerganov/llama.cpp).
52
+ * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI. Supports GGUF with GPU acceleration via the ctransformers backend - llama-cpp-python backend should work soon too.
53
+ * [KoboldCpp](https://github.com/LostRuins/koboldcpp), now supports GGUF as of release 1.41! A powerful GGML web UI, with full GPU accel. Especially good for story telling.
54
+ * [LM Studio](https://lmstudio.ai/), version 0.2.2 and later support GGUF. A fully featured local GUI with GPU acceleration on both Windows (NVidia and AMD), and macOS.
55
+ * [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), should now work, choose the `c_transformers` backend. A great web UI with many interesting features. Supports CUDA GPU acceleration.
56
+ * [ctransformers](https://github.com/marella/ctransformers), now supports GGUF as of version 0.2.24! A Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
57
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), supports GGUF as of version 0.1.79. A Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
58
+ * [candle](https://github.com/huggingface/candle), added GGUF support on August 22nd. Candle is a Rust ML framework with a focus on performance, including GPU support, and ease of use.
59
+
60
+ <!-- README_GGUF.md-about-gguf end -->
61
+ <!-- repositories-available start -->
62
+ ## Repositories available
63
+
64
+ * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/orca_mini_v3_70B-GPTQ)
65
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF)
66
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGML)
67
+ * [Pankaj Mathur's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/psmathur/orca_mini_v3_70b)
68
+ <!-- repositories-available end -->
69
+
70
+ <!-- prompt-template start -->
71
+ ## Prompt template: Orca-Hashes
72
+
73
+ ```
74
+ ### System:
75
+ {system_message}
76
+
77
+ ### User:
78
+ {prompt}
79
+
80
+ ### Assistant:
81
+
82
+ ```
83
+
84
+ <!-- prompt-template end -->
85
+ <!-- compatibility_gguf start -->
86
+ ## Compatibility
87
+
88
+ These quantised GGUF files are compatible with llama.cpp from August 21st 2023 onwards, as of commit [6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9](https://github.com/ggerganov/llama.cpp/commit/6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9)
89
+
90
+ They are now also compatible with many third party UIs and libraries - please see the list at the top of the README.
91
+
92
+ ## Explanation of quantisation methods
93
+ <details>
94
+ <summary>Click to see details</summary>
95
+
96
+ The new methods available are:
97
+ * 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)
98
+ * 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.
99
+ * 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.
100
+ * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
101
+ * 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
102
+
103
+ Refer to the Provided Files table below to see what files use which methods, and how.
104
+ </details>
105
+ <!-- compatibility_gguf end -->
106
+
107
+ <!-- README_GGUF.md-provided-files start -->
108
+ ## Provided files
109
+
110
+ | Name | Quant method | Bits | Size | Max RAM required | Use case |
111
+ | ---- | ---- | ---- | ---- | ---- | ----- |
112
+ | [orca_mini_v3_70b.Q6_K.gguf-split-b](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q6_K.gguf-split-b) | Q6_K | 6 | 19.89 GB| 22.39 GB | very large, extremely low quality loss |
113
+ | [orca_mini_v3_70b.Q2_K.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q2_K.gguf) | Q2_K | 2 | 29.28 GB| 31.78 GB | smallest, significant quality loss - not recommended for most purposes |
114
+ | [orca_mini_v3_70b.Q3_K_S.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q3_K_S.gguf) | Q3_K_S | 3 | 29.92 GB| 32.42 GB | very small, high quality loss |
115
+ | [orca_mini_v3_70b.Q3_K_M.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q3_K_M.gguf) | Q3_K_M | 3 | 33.19 GB| 35.69 GB | very small, high quality loss |
116
+ | [orca_mini_v3_70b.Q3_K_L.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q3_K_L.gguf) | Q3_K_L | 3 | 36.15 GB| 38.65 GB | small, substantial quality loss |
117
+ | [orca_mini_v3_70b.Q8_0.gguf-split-b](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q8_0.gguf-split-b) | Q8_0 | 8 | 36.59 GB| 39.09 GB | very large, extremely low quality loss - not recommended |
118
+ | [orca_mini_v3_70b.Q6_K.gguf-split-a](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q6_K.gguf-split-a) | Q6_K | 6 | 36.70 GB| 39.20 GB | very large, extremely low quality loss |
119
+ | [orca_mini_v3_70b.Q8_0.gguf-split-a](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q8_0.gguf-split-a) | Q8_0 | 8 | 36.70 GB| 39.20 GB | very large, extremely low quality loss - not recommended |
120
+ | [orca_mini_v3_70b.Q4_0.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q4_0.gguf) | Q4_0 | 4 | 38.87 GB| 41.37 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
121
+ | [orca_mini_v3_70b.Q4_K_S.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q4_K_S.gguf) | Q4_K_S | 4 | 39.07 GB| 41.57 GB | small, greater quality loss |
122
+ | [orca_mini_v3_70b.Q4_K_M.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q4_K_M.gguf) | Q4_K_M | 4 | 41.42 GB| 43.92 GB | medium, balanced quality - recommended |
123
+ | [orca_mini_v3_70b.Q5_0.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q5_0.gguf) | Q5_0 | 5 | 47.46 GB| 49.96 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
124
+ | [orca_mini_v3_70b.Q5_K_S.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q5_K_S.gguf) | Q5_K_S | 5 | 47.46 GB| 49.96 GB | large, low quality loss - recommended |
125
+ | [orca_mini_v3_70b.Q5_K_M.gguf](https://huggingface.co/TheBloke/orca_mini_v3_70B-GGUF/blob/main/orca_mini_v3_70b.Q5_K_M.gguf) | Q5_K_M | 5 | 48.75 GB| 51.25 GB | large, very low quality loss - recommended |
126
+ | orca_mini_v3_70b.Q6_K.gguf | Q6_K | 6 | 56.59 GB| 59.09 GB | very large, extremely low quality loss |
127
+ | orca_mini_v3_70b.Q8_0.gguf | Q8_0 | 8 | 73.29 GB| 75.79 GB | very large, extremely low quality loss - not recommended |
128
+
129
+ **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.
130
+
131
+ ### Q6_K and Q8_0 files are split and require joining
132
+
133
+ **Note:** HF does not support uploading files larger than 50GB. Therefore I have uploaded the Q6_K and Q8_0 files as split files.
134
+
135
+ <details>
136
+ <summary>Click for instructions regarding Q6_K and Q8_0 files</summary>
137
+
138
+ ### q6_K
139
+ Please download:
140
+ * `orca_mini_v3_70b.Q6_K.gguf-split-a`
141
+ * `orca_mini_v3_70b.Q6_K.gguf-split-b`
142
+
143
+ ### q8_0
144
+ Please download:
145
+ * `orca_mini_v3_70b.Q8_0.gguf-split-a`
146
+ * `orca_mini_v3_70b.Q8_0.gguf-split-b`
147
+
148
+ To join the files, do the following:
149
+
150
+ Linux and macOS:
151
+ ```
152
+ cat orca_mini_v3_70b.Q6_K.gguf-split-* > orca_mini_v3_70b.Q6_K.gguf && rm orca_mini_v3_70b.Q6_K.gguf-split-*
153
+ cat orca_mini_v3_70b.Q8_0.gguf-split-* > orca_mini_v3_70b.Q8_0.gguf && rm orca_mini_v3_70b.Q8_0.gguf-split-*
154
+ ```
155
+ Windows command line:
156
+ ```
157
+ COPY /B orca_mini_v3_70b.Q6_K.gguf-split-a + orca_mini_v3_70b.Q6_K.gguf-split-b orca_mini_v3_70b.Q6_K.gguf
158
+ del orca_mini_v3_70b.Q6_K.gguf-split-a orca_mini_v3_70b.Q6_K.gguf-split-b
159
+
160
+ COPY /B orca_mini_v3_70b.Q8_0.gguf-split-a + orca_mini_v3_70b.Q8_0.gguf-split-b orca_mini_v3_70b.Q8_0.gguf
161
+ del orca_mini_v3_70b.Q8_0.gguf-split-a orca_mini_v3_70b.Q8_0.gguf-split-b
162
+ ```
163
+
164
+ </details>
165
+ <!-- README_GGUF.md-provided-files end -->
166
+
167
+ <!-- README_GGUF.md-how-to-run start -->
168
+ ## Example `llama.cpp` command
169
+
170
+ Make sure you are using `llama.cpp` from commit [6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9](https://github.com/ggerganov/llama.cpp/commit/6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9) or later.
171
+
172
+ For compatibility with older versions of llama.cpp, or for any third-party libraries or clients that haven't yet updated for GGUF, please use GGML files instead.
173
+
174
+ ```
175
+ ./main -t 10 -ngl 32 -m orca_mini_v3_70b.q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### System:\nYou are a story writing assistant.\n\n### User:\nWrite a story about llamas\n\n### Assistant:"
176
+ ```
177
+ Change `-t 10` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`. If offloading all layers to GPU, set `-t 1`.
178
+
179
+ Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
180
+
181
+ Change `-c 4096` to the desired sequence length for this model. 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.
182
+
183
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
184
+
185
+ 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)
186
+
187
+ ## How to run in `text-generation-webui`
188
+
189
+ Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
190
+
191
+ ## How to run from Python code
192
+
193
+ 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.
194
+
195
+ ### How to load this model from Python using ctransformers
196
+
197
+ #### First install the package
198
+
199
+ ```bash
200
+ # Base ctransformers with no GPU acceleration
201
+ pip install ctransformers>=0.2.24
202
+ # Or with CUDA GPU acceleration
203
+ pip install ctransformers[cuda]>=0.2.24
204
+ # Or with ROCm GPU acceleration
205
+ CT_HIPBLAS=1 pip install ctransformers>=0.2.24 --no-binary ctransformers
206
+ # Or with Metal GPU acceleration for macOS systems
207
+ CT_METAL=1 pip install ctransformers>=0.2.24 --no-binary ctransformers
208
+ ```
209
+
210
+ #### Simple example code to load one of these GGUF models
211
+
212
+ ```python
213
+ from ctransformers import AutoModelForCausalLM
214
+
215
+ # Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
216
+ llm = AutoModelForCausalLM.from_pretrained("TheBloke/orca_mini_v3_70B-GGUF", model_file="orca_mini_v3_70b.q4_K_M.gguf", model_type="llama", gpu_layers=50)
217
+
218
+ print(llm("AI is going to"))
219
+ ```
220
+
221
+ ## How to use with LangChain
222
+
223
+ Here's guides on using llama-cpp-python or ctransformers with LangChain:
224
+
225
+ * [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
226
+ * [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
227
+
228
+ <!-- README_GGUF.md-how-to-run end -->
229
+
230
+ <!-- footer start -->
231
+ <!-- 200823 -->
232
+ ## Discord
233
+
234
+ For further support, and discussions on these models and AI in general, join us at:
235
+
236
+ [TheBloke AI's Discord server](https://discord.gg/theblokeai)
237
+
238
+ ## Thanks, and how to contribute.
239
+
240
+ Thanks to the [chirper.ai](https://chirper.ai) team!
241
+
242
+ 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.
243
+
244
+ 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.
245
+
246
+ Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
247
+
248
+ * Patreon: https://patreon.com/TheBlokeAI
249
+ * Ko-Fi: https://ko-fi.com/TheBlokeAI
250
+
251
+ **Special thanks to**: Aemon Algiz.
252
+
253
+ **Patreon special mentions**: Russ Johnson, J, alfie_i, Alex, NimbleBox.ai, Chadd, Mandus, Nikolai Manek, Ken Nordquist, ya boyyy, Illia Dulskyi, Viktor Bowallius, vamX, Iucharbius, zynix, Magnesian, Clay Pascal, Pierre Kircher, Enrico Ros, Tony Hughes, Elle, Andrey, knownsqashed, Deep Realms, Jerry Meng, Lone Striker, Derek Yates, Pyrater, Mesiah Bishop, James Bentley, Femi Adebogun, Brandon Frisco, SuperWojo, Alps Aficionado, Michael Dempsey, Vitor Caleffi, Will Dee, Edmond Seymore, usrbinkat, LangChain4j, Kacper Wikieł, Luke Pendergrass, John Detwiler, theTransient, Nathan LeClaire, Tiffany J. Kim, biorpg, Eugene Pentland, Stanislav Ovsiannikov, Fred von Graf, terasurfer, Kalila, Dan Guido, Nitin Borwankar, 阿明, Ai Maven, John Villwock, Gabriel Puliatti, Stephen Murray, Asp the Wyvern, danny, Chris Smitley, ReadyPlayerEmma, S_X, Daniel P. Andersen, Olakabola, Jeffrey Morgan, Imad Khwaja, Caitlyn Gatomon, webtim, Alicia Loh, Trenton Dambrowitz, Swaroop Kallakuri, Erik Bjäreholt, Leonard Tan, Spiking Neurons AB, Luke @flexchar, Ajan Kanaga, Thomas Belote, Deo Leter, RoA, Willem Michiel, transmissions 11, subjectnull, Matthew Berman, Joseph William Delisle, David Ziegler, Michael Davis, Johann-Peter Hartmann, Talal Aujan, senxiiz, Artur Olbinski, Rainer Wilmers, Spencer Kim, Fen Risland, Cap'n Zoog, Rishabh Srivastava, Michael Levine, Geoffrey Montalvo, Sean Connelly, Alexandros Triantafyllidis, Pieter, Gabriel Tamborski, Sam, Subspace Studios, Junyu Yang, Pedro Madruga, Vadim, Cory Kujawski, K, Raven Klaugh, Randy H, Mano Prime, Sebastain Graf, Space Cruiser
254
+
255
+
256
+ Thank you to all my generous patrons and donaters!
257
+
258
+ And thank you again to a16z for their generous grant.
259
+
260
+ <!-- footer end -->
261
+
262
+ <!-- original-model-card start -->
263
+ # Original model card: Pankaj Mathur's Orca Mini v3 70B
264
+
265
+
266
+ # orca_mini_v3_70b
267
+
268
+ A Llama2-70b model trained on Orca Style datasets.
269
+
270
+
271
+ <br>
272
+
273
+ ![orca-mini](https://huggingface.co/psmathur/orca_mini_v3_70b/resolve/main/orca_minis_small.jpeg)
274
+
275
+
276
+ <br>
277
+
278
+ **P.S. If you're interested to collaborate, please connect with me at www.linkedin.com/in/pankajam.**
279
+
280
+ <br>
281
+
282
+ ### quantized versions
283
+
284
+ Big thanks to [@TheBloke](https://huggingface.co/TheBloke)
285
+
286
+ 1) https://huggingface.co/TheBloke/orca_mini_v3_70B-GGML
287
+
288
+ 2) https://huggingface.co/TheBloke/orca_mini_v3_70B-GPTQ
289
+
290
+ <br>
291
+
292
+ #### license disclaimer:
293
+
294
+ This model is bound by the license & usage restrictions of the original Llama-2 model. And comes with no warranty or gurantees of any kind.
295
+
296
+ <br>
297
+
298
+ ## Evaluation
299
+
300
+ We evaluated orca_mini_v3_70b on a wide range of tasks using [Language Model Evaluation Harness](https://github.com/EleutherAI/lm-evaluation-harness) from EleutherAI.
301
+
302
+ Here are the results on metrics used by [HuggingFaceH4 Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
303
+
304
+ |||||
305
+ |:------:|:--------:|:-------:|:--------:|
306
+ |**Task**|**Metric**|**Value**|**Stderr**|
307
+ |*arc_challenge*|acc_norm|0.7098|0.0132|
308
+ |*hellaswag*|acc_norm|0.8779|0.0032|
309
+ |*mmlu*|acc_norm|0.6904|0.0351|
310
+ |*truthfulqa_mc*|mc2|0.6196|0.0151|
311
+ |**Total Average**|-|**0.722175**||
312
+
313
+
314
+ <br>
315
+
316
+ ## Example Usage
317
+
318
+ Here is the prompt format
319
+
320
+ ```
321
+ ### System:
322
+ You are an AI assistant that follows instruction extremely well. Help as much as you can.
323
+
324
+ ### User:
325
+ Tell me about Orcas.
326
+
327
+ ### Assistant:
328
+
329
+ ```
330
+
331
+ Below shows a code example on how to use this model
332
+
333
+ ```python
334
+ import torch
335
+ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
336
+
337
+ tokenizer = AutoTokenizer.from_pretrained("psmathur/orca_mini_v3_70b")
338
+ model = AutoModelForCausalLM.from_pretrained(
339
+ "psmathur/orca_mini_v3_70b",
340
+ torch_dtype=torch.float16,
341
+ load_in_8bit=True,
342
+ low_cpu_mem_usage=True,
343
+ device_map="auto"
344
+ )
345
+ system_prompt = "### System:\nYou are an AI assistant that follows instruction extremely well. Help as much as you can.\n\n"
346
+
347
+ #generate text steps
348
+ instruction = "Tell me about Orcas."
349
+ prompt = f"{system_prompt}### User: {instruction}\n\n### Assistant:\n"
350
+ inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
351
+ output = model.generate(**inputs, do_sample=True, top_p=0.95, top_k=0, max_new_tokens=4096)
352
+
353
+ print(tokenizer.decode(output[0], skip_special_tokens=True))
354
+
355
+ ```
356
+
357
+ <br>
358
+
359
+ #### Limitations & Biases:
360
+
361
+ While this model aims for accuracy, it can occasionally produce inaccurate or misleading results.
362
+
363
+ Despite diligent efforts in refining the pretraining data, there remains a possibility for the generation of inappropriate, biased, or offensive content.
364
+
365
+ Exercise caution and cross-check information when necessary.
366
+
367
+ <br>
368
+
369
+ ### Citiation:
370
+
371
+ Please kindly cite using the following BibTeX:
372
+
373
+ ```
374
+ @misc{orca_mini_v3_70b,
375
+ author = {Pankaj Mathur},
376
+ title = {orca_mini_v3_70b: An Orca Style Llama2-70b model},
377
+ year = {2023},
378
+ publisher = {HuggingFace},
379
+ journal = {HuggingFace repository},
380
+ howpublished = {\url{https://https://huggingface.co/psmathur/orca_mini_v3_70b},
381
+ }
382
+ ```
383
+
384
+ ```
385
+ @misc{mukherjee2023orca,
386
+ title={Orca: Progressive Learning from Complex Explanation Traces of GPT-4},
387
+ author={Subhabrata Mukherjee and Arindam Mitra and Ganesh Jawahar and Sahaj Agarwal and Hamid Palangi and Ahmed Awadallah},
388
+ year={2023},
389
+ eprint={2306.02707},
390
+ archivePrefix={arXiv},
391
+ primaryClass={cs.CL}
392
+ }
393
+ ```
394
+
395
+ ```
396
+ @software{touvron2023llama2,
397
+ title={Llama 2: Open Foundation and Fine-Tuned Chat Models},
398
+ author={Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava,
399
+ Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller,
400
+ Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann,
401
+ Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov,
402
+ Pushkar Mishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith,
403
+ Ranjan Subramanian, Xiaoqing Ellen Tan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu , Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan,
404
+ Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom},
405
+ year={2023}
406
+ }
407
+ ```
408
+
409
+ <!-- original-model-card end -->