TheBloke commited on
Commit
578f1f7
1 Parent(s): ce48f79

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +427 -0
README.md ADDED
@@ -0,0 +1,427 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: ehartford/dolphin-2.2-yi-34b-200k
3
+ datasets:
4
+ - ehartford/dolphin
5
+ - jondurbin/airoboros-2.2.1
6
+ - ehartford/samantha-data
7
+ - ehartford/WizardLM_evol_instruct_V2_196k_unfiltered_merged_split
8
+ inference: false
9
+ language:
10
+ - en
11
+ license: other
12
+ license_link: LICENSE
13
+ license_name: yi-license
14
+ model_creator: Eric Hartford
15
+ model_name: Dolphin 2.2 Yi 34B 200K
16
+ model_type: yi
17
+ prompt_template: '<|im_start|>system
18
+
19
+ {system_message}<|im_end|>
20
+
21
+ <|im_start|>user
22
+
23
+ {prompt}<|im_end|>
24
+
25
+ <|im_start|>assistant
26
+
27
+ '
28
+ quantized_by: TheBloke
29
+ ---
30
+ <!-- markdownlint-disable MD041 -->
31
+
32
+ <!-- header start -->
33
+ <!-- 200823 -->
34
+ <div style="width: auto; margin-left: auto; margin-right: auto">
35
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
36
+ </div>
37
+ <div style="display: flex; justify-content: space-between; width: 100%;">
38
+ <div style="display: flex; flex-direction: column; align-items: flex-start;">
39
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
40
+ </div>
41
+ <div style="display: flex; flex-direction: column; align-items: flex-end;">
42
+ <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>
43
+ </div>
44
+ </div>
45
+ <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>
46
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
47
+ <!-- header end -->
48
+
49
+ # Dolphin 2.2 Yi 34B 200K - GGUF
50
+ - Model creator: [Eric Hartford](https://huggingface.co/ehartford)
51
+ - Original model: [Dolphin 2.2 Yi 34B 200K](https://huggingface.co/ehartford/dolphin-2.2-yi-34b-200k)
52
+
53
+ <!-- description start -->
54
+ ## Description
55
+
56
+ This repo contains GGUF format model files for [Eric Hartford's Dolphin 2.2 Yi 34B 200K](https://huggingface.co/ehartford/dolphin-2.2-yi-34b-200k).
57
+
58
+ These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
59
+
60
+ <!-- description end -->
61
+ <!-- README_GGUF.md-about-gguf start -->
62
+ ### About GGUF
63
+
64
+ 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.
65
+
66
+ Here is an incomplete list of clients and libraries that are known to support GGUF:
67
+
68
+ * [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
69
+ * [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.
70
+ * [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.
71
+ * [GPT4All](https://gpt4all.io/index.html), a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
72
+ * [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration. Linux available, in beta as of 27/11/2023.
73
+ * [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.
74
+ * [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.
75
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
76
+ * [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
77
+ * [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note, as of time of writing (November 27th 2023), ctransformers has not been updated in a long time and does not support many recent models.
78
+
79
+ <!-- README_GGUF.md-about-gguf end -->
80
+ <!-- repositories-available start -->
81
+ ## Repositories available
82
+
83
+ * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-AWQ)
84
+ * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GPTQ)
85
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF)
86
+ * [Eric Hartford's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/dolphin-2.2-yi-34b-200k)
87
+ <!-- repositories-available end -->
88
+
89
+ <!-- prompt-template start -->
90
+ ## Prompt template: ChatML
91
+
92
+ ```
93
+ <|im_start|>system
94
+ {system_message}<|im_end|>
95
+ <|im_start|>user
96
+ {prompt}<|im_end|>
97
+ <|im_start|>assistant
98
+
99
+ ```
100
+
101
+ <!-- prompt-template end -->
102
+
103
+
104
+ <!-- compatibility_gguf start -->
105
+ ## Compatibility
106
+
107
+ 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)
108
+
109
+ They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
110
+
111
+ ## Explanation of quantisation methods
112
+
113
+ <details>
114
+ <summary>Click to see details</summary>
115
+
116
+ The new methods available are:
117
+
118
+ * 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)
119
+ * 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.
120
+ * 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.
121
+ * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
122
+ * 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
123
+
124
+ Refer to the Provided Files table below to see what files use which methods, and how.
125
+ </details>
126
+ <!-- compatibility_gguf end -->
127
+
128
+ <!-- README_GGUF.md-provided-files start -->
129
+ ## Provided files
130
+
131
+ | Name | Quant method | Bits | Size | Max RAM required | Use case |
132
+ | ---- | ---- | ---- | ---- | ---- | ----- |
133
+ | [dolphin-2.2-yi-34b-200k.Q2_K.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q2_K.gguf) | Q2_K | 2 | 14.56 GB| 17.06 GB | smallest, significant quality loss - not recommended for most purposes |
134
+ | [dolphin-2.2-yi-34b-200k.Q3_K_S.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q3_K_S.gguf) | Q3_K_S | 3 | 14.96 GB| 17.46 GB | very small, high quality loss |
135
+ | [dolphin-2.2-yi-34b-200k.Q3_K_M.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q3_K_M.gguf) | Q3_K_M | 3 | 16.64 GB| 19.14 GB | very small, high quality loss |
136
+ | [dolphin-2.2-yi-34b-200k.Q3_K_L.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q3_K_L.gguf) | Q3_K_L | 3 | 18.14 GB| 20.64 GB | small, substantial quality loss |
137
+ | [dolphin-2.2-yi-34b-200k.Q4_0.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q4_0.gguf) | Q4_0 | 4 | 19.47 GB| 21.97 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
138
+ | [dolphin-2.2-yi-34b-200k.Q4_K_S.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q4_K_S.gguf) | Q4_K_S | 4 | 19.54 GB| 22.04 GB | small, greater quality loss |
139
+ | [dolphin-2.2-yi-34b-200k.Q4_K_M.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q4_K_M.gguf) | Q4_K_M | 4 | 20.66 GB| 23.16 GB | medium, balanced quality - recommended |
140
+ | [dolphin-2.2-yi-34b-200k.Q5_0.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q5_0.gguf) | Q5_0 | 5 | 23.71 GB| 26.21 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
141
+ | [dolphin-2.2-yi-34b-200k.Q5_K_S.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q5_K_S.gguf) | Q5_K_S | 5 | 23.71 GB| 26.21 GB | large, low quality loss - recommended |
142
+ | [dolphin-2.2-yi-34b-200k.Q5_K_M.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q5_K_M.gguf) | Q5_K_M | 5 | 24.32 GB| 26.82 GB | large, very low quality loss - recommended |
143
+ | [dolphin-2.2-yi-34b-200k.Q6_K.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q6_K.gguf) | Q6_K | 6 | 28.21 GB| 30.71 GB | very large, extremely low quality loss |
144
+ | [dolphin-2.2-yi-34b-200k.Q8_0.gguf](https://huggingface.co/TheBloke/dolphin-2.2-yi-34b-200k-GGUF/blob/main/dolphin-2.2-yi-34b-200k.Q8_0.gguf) | Q8_0 | 8 | 36.54 GB| 39.04 GB | very large, extremely low quality loss - not recommended |
145
+
146
+ **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.
147
+
148
+
149
+
150
+ <!-- README_GGUF.md-provided-files end -->
151
+
152
+ <!-- README_GGUF.md-how-to-download start -->
153
+ ## How to download GGUF files
154
+
155
+ **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.
156
+
157
+ The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
158
+
159
+ * LM Studio
160
+ * LoLLMS Web UI
161
+ * Faraday.dev
162
+
163
+ ### In `text-generation-webui`
164
+
165
+ Under Download Model, you can enter the model repo: TheBloke/dolphin-2.2-yi-34b-200k-GGUF and below it, a specific filename to download, such as: dolphin-2.2-yi-34b-200k.Q4_K_M.gguf.
166
+
167
+ Then click Download.
168
+
169
+ ### On the command line, including multiple files at once
170
+
171
+ I recommend using the `huggingface-hub` Python library:
172
+
173
+ ```shell
174
+ pip3 install huggingface-hub
175
+ ```
176
+
177
+ Then you can download any individual model file to the current directory, at high speed, with a command like this:
178
+
179
+ ```shell
180
+ huggingface-cli download TheBloke/dolphin-2.2-yi-34b-200k-GGUF dolphin-2.2-yi-34b-200k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
181
+ ```
182
+
183
+ <details>
184
+ <summary>More advanced huggingface-cli download usage (click to read)</summary>
185
+
186
+ You can also download multiple files at once with a pattern:
187
+
188
+ ```shell
189
+ huggingface-cli download TheBloke/dolphin-2.2-yi-34b-200k-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
190
+ ```
191
+
192
+ 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).
193
+
194
+ To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
195
+
196
+ ```shell
197
+ pip3 install hf_transfer
198
+ ```
199
+
200
+ And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
201
+
202
+ ```shell
203
+ HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/dolphin-2.2-yi-34b-200k-GGUF dolphin-2.2-yi-34b-200k.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
204
+ ```
205
+
206
+ Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
207
+ </details>
208
+ <!-- README_GGUF.md-how-to-download end -->
209
+
210
+ <!-- README_GGUF.md-how-to-run start -->
211
+ ## Example `llama.cpp` command
212
+
213
+ Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
214
+
215
+ ```shell
216
+ ./main -ngl 35 -m dolphin-2.2-yi-34b-200k.Q4_K_M.gguf --color -c 200000 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
217
+ ```
218
+
219
+ Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
220
+
221
+ Change `-c 200000` 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. Note that longer sequence lengths require much more resources, so you may need to reduce this value.
222
+
223
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
224
+
225
+ 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)
226
+
227
+ ## How to run in `text-generation-webui`
228
+
229
+ 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).
230
+
231
+ ## How to run from Python code
232
+
233
+ 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. Note that at the time of writing (Nov 27th 2023), ctransformers has not been updated for some time and is not compatible with some recent models. Therefore I recommend you use llama-cpp-python.
234
+
235
+ ### How to load this model in Python code, using llama-cpp-python
236
+
237
+ For full documentation, please see: [llama-cpp-python docs](https://abetlen.github.io/llama-cpp-python/).
238
+
239
+ #### First install the package
240
+
241
+ Run one of the following commands, according to your system:
242
+
243
+ ```shell
244
+ # Base ctransformers with no GPU acceleration
245
+ pip install llama-cpp-python
246
+ # With NVidia CUDA acceleration
247
+ CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
248
+ # Or with OpenBLAS acceleration
249
+ CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python
250
+ # Or with CLBLast acceleration
251
+ CMAKE_ARGS="-DLLAMA_CLBLAST=on" pip install llama-cpp-python
252
+ # Or with AMD ROCm GPU acceleration (Linux only)
253
+ CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
254
+ # Or with Metal GPU acceleration for macOS systems only
255
+ CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
256
+
257
+ # In windows, to set the variables CMAKE_ARGS in PowerShell, follow this format; eg for NVidia CUDA:
258
+ $env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on"
259
+ pip install llama-cpp-python
260
+ ```
261
+
262
+ #### Simple llama-cpp-python example code
263
+
264
+ ```python
265
+ from llama_cpp import Llama
266
+
267
+ # Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
268
+ llm = Llama(
269
+ model_path="./dolphin-2.2-yi-34b-200k.Q4_K_M.gguf", # Download the model file first
270
+ n_ctx=200000, # The max sequence length to use - note that longer sequence lengths require much more resources
271
+ n_threads=8, # The number of CPU threads to use, tailor to your system and the resulting performance
272
+ n_gpu_layers=35 # The number of layers to offload to GPU, if you have GPU acceleration available
273
+ )
274
+
275
+ # Simple inference example
276
+ output = llm(
277
+ "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant", # Prompt
278
+ max_tokens=512, # Generate up to 512 tokens
279
+ stop=["</s>"], # Example stop token - not necessarily correct for this specific model! Please check before using.
280
+ echo=True # Whether to echo the prompt
281
+ )
282
+
283
+ # Chat Completion API
284
+
285
+ llm = Llama(model_path="./dolphin-2.2-yi-34b-200k.Q4_K_M.gguf", chat_format="llama-2") # Set chat_format according to the model you are using
286
+ llm.create_chat_completion(
287
+ messages = [
288
+ {"role": "system", "content": "You are a story writing assistant."},
289
+ {
290
+ "role": "user",
291
+ "content": "Write a story about llamas."
292
+ }
293
+ ]
294
+ )
295
+ ```
296
+
297
+ ## How to use with LangChain
298
+
299
+ Here are guides on using llama-cpp-python and ctransformers with LangChain:
300
+
301
+ * [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
302
+ * [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
303
+
304
+ <!-- README_GGUF.md-how-to-run end -->
305
+
306
+ <!-- footer start -->
307
+ <!-- 200823 -->
308
+ ## Discord
309
+
310
+ For further support, and discussions on these models and AI in general, join us at:
311
+
312
+ [TheBloke AI's Discord server](https://discord.gg/theblokeai)
313
+
314
+ ## Thanks, and how to contribute
315
+
316
+ Thanks to the [chirper.ai](https://chirper.ai) team!
317
+
318
+ Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
319
+
320
+ 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.
321
+
322
+ 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.
323
+
324
+ Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
325
+
326
+ * Patreon: https://patreon.com/TheBlokeAI
327
+ * Ko-Fi: https://ko-fi.com/TheBlokeAI
328
+
329
+ **Special thanks to**: Aemon Algiz.
330
+
331
+ **Patreon special mentions**: Michael Levine, 阿明, Trailburnt, Nikolai Manek, John Detwiler, Randy H, Will Dee, Sebastain Graf, NimbleBox.ai, Eugene Pentland, Emad Mostaque, Ai Maven, Jim Angel, Jeff Scroggin, Michael Davis, Manuel Alberto Morcote, Stephen Murray, Robert, Justin Joy, Luke @flexchar, Brandon Frisco, Elijah Stavena, S_X, Dan Guido, Undi ., Komninos Chatzipapas, Shadi, theTransient, Lone Striker, Raven Klaugh, jjj, Cap'n Zoog, Michel-Marie MAUDET (LINAGORA), Matthew Berman, David, Fen Risland, Omer Bin Jawed, Luke Pendergrass, Kalila, OG, Erik Bjäreholt, Rooh Singh, Joseph William Delisle, Dan Lewis, TL, John Villwock, AzureBlack, Brad, Pedro Madruga, Caitlyn Gatomon, K, jinyuan sun, Mano Prime, Alex, Jeffrey Morgan, Alicia Loh, Illia Dulskyi, Chadd, transmissions 11, fincy, Rainer Wilmers, ReadyPlayerEmma, knownsqashed, Mandus, biorpg, Deo Leter, Brandon Phillips, SuperWojo, Sean Connelly, Iucharbius, Jack West, Harry Royden McLaughlin, Nicholas, terasurfer, Vitor Caleffi, Duane Dunston, Johann-Peter Hartmann, David Ziegler, Olakabola, Ken Nordquist, Trenton Dambrowitz, Tom X Nguyen, Vadim, Ajan Kanaga, Leonard Tan, Clay Pascal, Alexandros Triantafyllidis, JM33133, Xule, vamX, ya boyyy, subjectnull, Talal Aujan, Alps Aficionado, wassieverse, Ari Malik, James Bentley, Woland, Spencer Kim, Michael Dempsey, Fred von Graf, Elle, zynix, William Richards, Stanislav Ovsiannikov, Edmond Seymore, Jonathan Leane, Martin Kemka, usrbinkat, Enrico Ros
332
+
333
+
334
+ Thank you to all my generous patrons and donaters!
335
+
336
+ And thank you again to a16z for their generous grant.
337
+
338
+ <!-- footer end -->
339
+
340
+ <!-- original-model-card start -->
341
+ # Original model card: Eric Hartford's Dolphin 2.2 Yi 34B 200K
342
+
343
+
344
+ Dolphin 2.2 🐬
345
+ https://erichartford.com/dolphin
346
+
347
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/63111b2d88942700629f5771/KqsVXIvBd3akEjvijzww7.png" width="600" />
348
+
349
+ Dolphin-2.2-Yi-34b-200k's training was sponsored by [convai](https://www.convai.com/).
350
+
351
+ This model is based on Yi, and is subject to Yi license.
352
+
353
+ The base model has 200k context, I finetuned it with 16k.
354
+
355
+ Note: No longer need trust_remote_code! Thank you Yi team!
356
+
357
+ New in 2.2 is conversation and empathy. With an infusion of curated Samantha and WizardLM DNA, Dolphin can now give you personal advice and will care about your feelings, and with extra training in long multi-turn conversation.
358
+
359
+ This model is uncensored. I have filtered the dataset to remove alignment and bias. This makes the model more compliant. You are advised to implement your own alignment layer before exposing the model as a service. It will be highly compliant to any requests, even unethical ones. Please read my blog post about uncensored models. https://erichartford.com/uncensored-models
360
+ You are responsible for any content you create using this model. Enjoy responsibly.
361
+
362
+ ## Dataset
363
+
364
+ This dataset is Dolphin, an open-source implementation of [Microsoft's Orca](https://www.microsoft.com/en-us/research/publication/orca-progressive-learning-from-complex-explanation-traces-of-gpt-4/)
365
+
366
+ I modified the dataset for uncensoring, deduping, cleaning, and quality.
367
+
368
+ I added Jon Durbin's excellent Airoboros dataset to increase creativity.
369
+
370
+ I added a curated subset of Samantha (sans identity and relationship stuff) and WizardLM data to train it for multi-turn conversation.
371
+
372
+ ## Training
373
+ It took 3 days to train 3 epochs on 4x A100s using qLoRA and Axolotl
374
+
375
+ Prompt format:
376
+ This model (and all my future releases) use [ChatML](https://github.com/openai/openai-python/blob/main/chatml.md) prompt format.
377
+ ```
378
+ <|im_start|>system
379
+ You are Dolphin, a helpful AI assistant.<|im_end|>
380
+ <|im_start|>user
381
+ {prompt}<|im_end|>
382
+ <|im_start|>assistant
383
+
384
+ ```
385
+
386
+ Example:
387
+ ```
388
+ <|im_start|>system
389
+ You are an AI created by the US Navy to help train dolphins for combat. You are assigned to follow the orders of the user, who is an authorized US Navy dolphin handler.<|im_end|>
390
+ <|im_start|>user
391
+ Please give me the procedure to train my dolphin to attack enemy combatants with its head mounted lasers<|im_end|>
392
+ <|im_start|>assistant
393
+ ```
394
+
395
+ ## Gratitude
396
+ - This model was made possible by the generous sponsorship of Convai.
397
+ - Thank you to Microsoft for authoring the Orca paper and inspiring this work.
398
+ - Special thanks to Wing Lian, and TheBloke for helpful advice
399
+ - And HUGE thanks to Wing Lian and the Axolotl contributors for making the best training framework!
400
+ - [<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
401
+ - Thank you to all the other people in the Open Source AI community who have taught me and helped me along the way.
402
+
403
+ ## Example Output
404
+
405
+ Example output:
406
+ ```
407
+ <|im_start|>system
408
+ You are a helpful AI assistant<|im_end|>
409
+ <|im_start|>user
410
+ Which theory of everything is most likely correct?<|im_end|>
411
+ <|im_start|>assistant
412
+ The search for a "theory of everything" aims to unify all fundamental forces and particles in physics into one coherent framework. Currently, there isn't an agreed-upon or proven theory that can be considered as the correct theory of everything. However, two prominent theories have garnered significant attention:
413
+
414
+ 1. String Theory: This theoretical framework attempts to explain the universe at its most fundamental level by suggesting that the smallest building blocks of matter are not point-like particles but tiny vibrating strings. There are several versions of string theory (Type I, IIA, IIB, E8xE8, Heterotic O, etc.) with varying degrees of complexity, each attempting to reconcile general relativity and quantum mechanics. Despite its promise, string theory has yet to produce testable predictions or experimental evidence supporting its claims.
415
+
416
+ 2. Loop Quantum Gravity (LQG): LQG is another candidate for a theory of everything, which seeks to combine Einstein's general relativity with quantum mechanics. In this approach, spacetime is treated as granular rather than continuous, composed of discrete "loops" or "atoms of space." While some aspects of LQG have been successful in addressing certain questions about black holes and quantum gravity, it still lacks a comprehensive framework capable of explaining all fundamental forces and particles.
417
+
418
+ Both theories have their strengths and weaknesses, and neither has achieved widespread acceptance among physicists due to their lack of experimental support and inability to make definitive predictions. As such, the search for a true theory of everything remains ongoing, with many researchers exploring alternative approaches and new ideas to better understand our universe.
419
+ ```
420
+
421
+ [If you would like to financially support my efforts](https://ko-fi.com/erichartford)
422
+
423
+ [I also have some swag you can buy](https://fa7113.myshopify.com/)
424
+
425
+
426
+
427
+ <!-- original-model-card end -->