MaziyarPanahi commited on
Commit
d940617
1 Parent(s): 27cd924

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -33,3 +33,13 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ OpenHermes-2.5-Mistral-7B.Q2_K.ggufm filter=lfs diff=lfs merge=lfs -text
37
+ OpenHermes-2.5-Mistral-7B.Q3_K_L.ggufm filter=lfs diff=lfs merge=lfs -text
38
+ OpenHermes-2.5-Mistral-7B.Q3_K_M.ggufm filter=lfs diff=lfs merge=lfs -text
39
+ OpenHermes-2.5-Mistral-7B.Q3_K_S.ggufm filter=lfs diff=lfs merge=lfs -text
40
+ OpenHermes-2.5-Mistral-7B.Q4_K_M.ggufm filter=lfs diff=lfs merge=lfs -text
41
+ OpenHermes-2.5-Mistral-7B.Q4_K_S.ggufm filter=lfs diff=lfs merge=lfs -text
42
+ OpenHermes-2.5-Mistral-7B.Q5_K_M.ggufm filter=lfs diff=lfs merge=lfs -text
43
+ OpenHermes-2.5-Mistral-7B.Q5_K_S.ggufm filter=lfs diff=lfs merge=lfs -text
44
+ OpenHermes-2.5-Mistral-7B.Q6_K.ggufm filter=lfs diff=lfs merge=lfs -text
45
+ OpenHermes-2.5-Mistral-7B.Q8_0.ggufm filter=lfs diff=lfs merge=lfs -text
OpenHermes-2.5-Mistral-7B.Q2_K.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e49b0e56b1a936e0f5aaee980026af0513614c6afa61655364a0f14c271369e3
3
+ size 2719251584
OpenHermes-2.5-Mistral-7B.Q3_K_L.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d09cc82f2a4fbefa93f4e74f54eaea35c73393bbf5f32e107a36f1742bbbc90
3
+ size 3822034880
OpenHermes-2.5-Mistral-7B.Q3_K_M.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d36f09a284185fbcd892dcbafcaa1907ed34188529f8c5eb2a85dc450229a1c
3
+ size 3518996416
OpenHermes-2.5-Mistral-7B.Q3_K_S.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82432ec5ed248eee91e822aa8adaf9ebdd6bc0b1ccc23ea5ad4252b1bcf4031f
3
+ size 3164577728
OpenHermes-2.5-Mistral-7B.Q4_K_M.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a073251ac83fb8c16a72664ae7b16fec17cfd10d492e67f395ca6ef72a677e87
3
+ size 4368450560
OpenHermes-2.5-Mistral-7B.Q4_K_S.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3e039ac246fce93375d0e9008f0416f2175403f1e6921c79786780963b8bfa3
3
+ size 4140385280
OpenHermes-2.5-Mistral-7B.Q5_K_M.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e68d28908e8049e8b082de5a7cf5b94367d1dc7c7de00cbe0271b064a0e605e
3
+ size 5131421696
OpenHermes-2.5-Mistral-7B.Q5_K_S.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2225b8f89e7c6efe5f2e38ec012986eb1b2e119777bd788ade44e0c913f998b
3
+ size 4997728256
OpenHermes-2.5-Mistral-7B.Q6_K.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5aaeac6df6adedafd2d4eb91dcd20dc6ec51303a799fc874febcc4a7c8af184
3
+ size 5942078528
OpenHermes-2.5-Mistral-7B.Q8_0.ggufm ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a25ee57419a6a591e93d01818092f81db8771b8d883a58f0c2043220973692be
3
+ size 7695875008
README.md ADDED
@@ -0,0 +1,233 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - quantized
5
+ - 2-bit
6
+ - 3-bit
7
+ - 4-bit
8
+ - 5-bit
9
+ - 6-bit
10
+ - 8-bit
11
+ - GGUF
12
+ - transformers
13
+ - pytorch
14
+ - safetensors
15
+ - mistral
16
+ - text-generation
17
+ - instruct
18
+ - finetune
19
+ - chatml
20
+ - gpt4
21
+ - synthetic data
22
+ - distillation
23
+ - en
24
+ - base_model:mistralai/Mistral-7B-v0.1
25
+ - license:apache-2.0
26
+ - autotrain_compatible
27
+ - endpoints_compatible
28
+ - has_space
29
+ - text-generation-inference
30
+ - region:us
31
+ model_name: OpenHermes-2.5-Mistral-7B-GGUF
32
+ base_model: teknium/OpenHermes-2.5-Mistral-7B
33
+ inference: false
34
+ model_creator: teknium
35
+ pipeline_tag: text-generation
36
+ quantized_by: MaziyarPanahi
37
+ ---
38
+ # [MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF](https://huggingface.co/MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF)
39
+ - Model creator: [teknium](https://huggingface.co/teknium)
40
+ - Original model: [teknium/OpenHermes-2.5-Mistral-7B](https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B)
41
+
42
+ ## Description
43
+ [MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF](https://huggingface.co/MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF) contains GGUF format model files for [teknium/OpenHermes-2.5-Mistral-7B](https://huggingface.co/teknium/OpenHermes-2.5-Mistral-7B).
44
+
45
+ ## How to use
46
+ Thanks to [TheBloke](https://huggingface.co/TheBloke) for preparing an amazing README on how to use GGUF models:
47
+
48
+ ### About GGUF
49
+
50
+ 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.
51
+
52
+ Here is an incomplete list of clients and libraries that are known to support GGUF:
53
+
54
+ * [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
55
+ * [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.
56
+ * [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.
57
+ * [GPT4All](https://gpt4all.io/index.html), a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
58
+ * [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.
59
+ * [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.
60
+ * [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.
61
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
62
+ * [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
63
+ * [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.
64
+
65
+ ### Explanation of quantisation methods
66
+
67
+ <details>
68
+ <summary>Click to see details</summary>
69
+
70
+ The new methods available are:
71
+
72
+ * 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)
73
+ * 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.
74
+ * 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.
75
+ * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
76
+ * 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
77
+
78
+ ## How to download GGUF files
79
+
80
+ **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.
81
+
82
+ The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
83
+
84
+ * LM Studio
85
+ * LoLLMS Web UI
86
+ * Faraday.dev
87
+
88
+ ### In `text-generation-webui`
89
+
90
+ Under Download Model, you can enter the model repo: [MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF](https://huggingface.co/MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF) and below it, a specific filename to download, such as: OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf.
91
+
92
+ Then click Download.
93
+
94
+ ### On the command line, including multiple files at once
95
+
96
+ I recommend using the `huggingface-hub` Python library:
97
+
98
+ ```shell
99
+ pip3 install huggingface-hub
100
+ ```
101
+
102
+ Then you can download any individual model file to the current directory, at high speed, with a command like this:
103
+
104
+ ```shell
105
+ huggingface-cli download MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
106
+ ```
107
+ </details>
108
+ <details>
109
+ <summary>More advanced huggingface-cli download usage (click to read)</summary>
110
+
111
+ You can also download multiple files at once with a pattern:
112
+
113
+ ```shell
114
+ huggingface-cli download [MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF](https://huggingface.co/MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF) --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
115
+ ```
116
+
117
+ 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).
118
+
119
+ To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
120
+
121
+ ```shell
122
+ pip3 install hf_transfer
123
+ ```
124
+
125
+ And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
126
+
127
+ ```shell
128
+ HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download MaziyarPanahi/OpenHermes-2.5-Mistral-7B-GGUF OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
129
+ ```
130
+
131
+ Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
132
+ </details>
133
+
134
+ ## Example `llama.cpp` command
135
+
136
+ Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
137
+
138
+ ```shell
139
+ ./main -ngl 35 -m OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf --color -c 32768 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system
140
+ {system_message}<|im_end|>
141
+ <|im_start|>user
142
+ {prompt}<|im_end|>
143
+ <|im_start|>assistant"
144
+ ```
145
+
146
+ Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
147
+
148
+ Change `-c 32768` 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.
149
+
150
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
151
+
152
+ 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)
153
+
154
+ ## How to run in `text-generation-webui`
155
+
156
+ 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).
157
+
158
+ ## How to run from Python code
159
+
160
+ 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.
161
+
162
+ ### How to load this model in Python code, using llama-cpp-python
163
+
164
+ For full documentation, please see: [llama-cpp-python docs](https://abetlen.github.io/llama-cpp-python/).
165
+
166
+ #### First install the package
167
+
168
+ Run one of the following commands, according to your system:
169
+
170
+ ```shell
171
+ # Base ctransformers with no GPU acceleration
172
+ pip install llama-cpp-python
173
+ # With NVidia CUDA acceleration
174
+ CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
175
+ # Or with OpenBLAS acceleration
176
+ CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python
177
+ # Or with CLBLast acceleration
178
+ CMAKE_ARGS="-DLLAMA_CLBLAST=on" pip install llama-cpp-python
179
+ # Or with AMD ROCm GPU acceleration (Linux only)
180
+ CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
181
+ # Or with Metal GPU acceleration for macOS systems only
182
+ CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
183
+
184
+ # In windows, to set the variables CMAKE_ARGS in PowerShell, follow this format; eg for NVidia CUDA:
185
+ $env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on"
186
+ pip install llama-cpp-python
187
+ ```
188
+
189
+ #### Simple llama-cpp-python example code
190
+
191
+ ```python
192
+ from llama_cpp import Llama
193
+
194
+ # Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
195
+ llm = Llama(
196
+ model_path="./OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf", # Download the model file first
197
+ n_ctx=32768, # The max sequence length to use - note that longer sequence lengths require much more resources
198
+ n_threads=8, # The number of CPU threads to use, tailor to your system and the resulting performance
199
+ n_gpu_layers=35 # The number of layers to offload to GPU, if you have GPU acceleration available
200
+ )
201
+
202
+ # Simple inference example
203
+ output = llm(
204
+ "<|im_start|>system
205
+ {system_message}<|im_end|>
206
+ <|im_start|>user
207
+ {prompt}<|im_end|>
208
+ <|im_start|>assistant", # Prompt
209
+ max_tokens=512, # Generate up to 512 tokens
210
+ stop=["</s>"], # Example stop token - not necessarily correct for this specific model! Please check before using.
211
+ echo=True # Whether to echo the prompt
212
+ )
213
+
214
+ # Chat Completion API
215
+
216
+ llm = Llama(model_path="./OpenHermes-2.5-Mistral-7B-GGUF.Q4_K_M.gguf", chat_format="llama-2") # Set chat_format according to the model you are using
217
+ llm.create_chat_completion(
218
+ messages = [
219
+ {"role": "system", "content": "You are a story writing assistant."},
220
+ {
221
+ "role": "user",
222
+ "content": "Write a story about llamas."
223
+ }
224
+ ]
225
+ )
226
+ ```
227
+
228
+ ## How to use with LangChain
229
+
230
+ Here are guides on using llama-cpp-python and ctransformers with LangChain:
231
+
232
+ * [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
233
+ * [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)