MaziyarPanahi commited on
Commit
966179e
1 Parent(s): 6c4d895

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -184
README.md CHANGED
@@ -25,192 +25,13 @@ quantized_by: MaziyarPanahi
25
  ## Description
26
  [MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF) contains GGUF format model files for [MaziyarPanahi/Llama-3-13B-Instruct-v0.1](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1).
27
 
28
- ## How to use
29
- Thanks to [TheBloke](https://huggingface.co/TheBloke) for preparing an amazing README on how to use GGUF models:
30
 
31
- ### About GGUF
32
 
33
- 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.
34
 
35
- Here is an incomplete list of clients and libraries that are known to support GGUF:
36
 
37
- * [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
38
- * [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.
39
- * [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.
40
- * [GPT4All](https://gpt4all.io/index.html), a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
41
- * [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.
42
- * [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.
43
- * [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.
44
- * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
45
- * [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
46
- * [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.
47
-
48
- ### Explanation of quantisation methods
49
-
50
- <details>
51
- <summary>Click to see details</summary>
52
-
53
- The new methods available are:
54
-
55
- * 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)
56
- * 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.
57
- * 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.
58
- * GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
59
- * 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
60
-
61
- ## How to download GGUF files
62
-
63
- **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.
64
-
65
- The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
66
-
67
- * LM Studio
68
- * LoLLMS Web UI
69
- * Faraday.dev
70
-
71
- ### In `text-generation-webui`
72
-
73
- Under Download Model, you can enter the model repo: [MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF) and below it, a specific filename to download, such as: Llama-3-13B-Instruct-v0.1-GGUF.Q4_K_M.gguf.
74
-
75
- Then click Download.
76
-
77
- ### On the command line, including multiple files at once
78
-
79
- I recommend using the `huggingface-hub` Python library:
80
-
81
- ```shell
82
- pip3 install huggingface-hub
83
  ```
84
-
85
- Then you can download any individual model file to the current directory, at high speed, with a command like this:
86
-
87
- ```shell
88
- huggingface-cli download MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF Llama-3-13B-Instruct-v0.1.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
89
- ```
90
- </details>
91
- <details>
92
- <summary>More advanced huggingface-cli download usage (click to read)</summary>
93
-
94
- You can also download multiple files at once with a pattern:
95
-
96
- ```shell
97
- huggingface-cli download [MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF) --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
98
- ```
99
-
100
- 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).
101
-
102
- To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
103
-
104
- ```shell
105
- pip3 install hf_transfer
106
- ```
107
-
108
- And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
109
-
110
- ```shell
111
- HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF Llama-3-13B-Instruct-v0.1.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
112
- ```
113
-
114
- Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
115
- </details>
116
-
117
- ## Example `llama.cpp` command
118
-
119
- Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
120
-
121
- ```shell
122
- ./main -ngl 35 -m Llama-3-13B-Instruct-v0.1.Q4_K_M.gguf --color -c 32768 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system
123
- {system_message}<|im_end|>
124
- <|im_start|>user
125
- {prompt}<|im_end|>
126
- <|im_start|>assistant"
127
- ```
128
-
129
- Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
130
-
131
- 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.
132
-
133
- If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
134
-
135
- 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)
136
-
137
- ## How to run in `text-generation-webui`
138
-
139
- 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-%20Model%20Tab.md#llamacpp).
140
-
141
- ## How to run from Python code
142
-
143
- 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.
144
-
145
- ### How to load this model in Python code, using llama-cpp-python
146
-
147
- For full documentation, please see: [llama-cpp-python docs](https://github.com/abetlen/llama-cpp-python/).
148
-
149
- #### First install the package
150
-
151
- Run one of the following commands, according to your system:
152
-
153
- ```shell
154
- # Base ctransformers with no GPU acceleration
155
- pip install llama-cpp-python
156
- # With NVidia CUDA acceleration
157
- CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
158
- # Or with OpenBLAS acceleration
159
- CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python
160
- # Or with CLBLast acceleration
161
- CMAKE_ARGS="-DLLAMA_CLBLAST=on" pip install llama-cpp-python
162
- # Or with AMD ROCm GPU acceleration (Linux only)
163
- CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
164
- # Or with Metal GPU acceleration for macOS systems only
165
- CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
166
-
167
- # In windows, to set the variables CMAKE_ARGS in PowerShell, follow this format; eg for NVidia CUDA:
168
- $env:CMAKE_ARGS = "-DLLAMA_OPENBLAS=on"
169
- pip install llama-cpp-python
170
- ```
171
-
172
- #### Simple llama-cpp-python example code
173
-
174
- ```python
175
- from llama_cpp import Llama
176
-
177
- # Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
178
- llm = Llama(
179
- model_path="./Llama-3-13B-Instruct-v0.1.Q4_K_M.gguf", # Download the model file first
180
- n_ctx=32768, # The max sequence length to use - note that longer sequence lengths require much more resources
181
- n_threads=8, # The number of CPU threads to use, tailor to your system and the resulting performance
182
- n_gpu_layers=35 # The number of layers to offload to GPU, if you have GPU acceleration available
183
- )
184
-
185
- # Simple inference example
186
- output = llm(
187
- "<|im_start|>system
188
- {system_message}<|im_end|>
189
- <|im_start|>user
190
- {prompt}<|im_end|>
191
- <|im_start|>assistant", # Prompt
192
- max_tokens=512, # Generate up to 512 tokens
193
- stop=["</s>"], # Example stop token - not necessarily correct for this specific model! Please check before using.
194
- echo=True # Whether to echo the prompt
195
- )
196
-
197
- # Chat Completion API
198
-
199
- llm = Llama(model_path="./Llama-3-13B-Instruct-v0.1.Q4_K_M.gguf", chat_format="llama-2") # Set chat_format according to the model you are using
200
- llm.create_chat_completion(
201
- messages = [
202
- {"role": "system", "content": "You are a story writing assistant."},
203
- {
204
- "role": "user",
205
- "content": "Write a story about llamas."
206
- }
207
- ]
208
- )
209
- ```
210
-
211
- ## How to use with LangChain
212
-
213
- Here are guides on using llama-cpp-python and ctransformers with LangChain:
214
-
215
- * [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
216
- * [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
 
25
  ## Description
26
  [MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1-GGUF) contains GGUF format model files for [MaziyarPanahi/Llama-3-13B-Instruct-v0.1](https://huggingface.co/MaziyarPanahi/Llama-3-13B-Instruct-v0.1).
27
 
 
 
28
 
29
+ ## Load GGUF models
30
 
31
+ You `MUST` follow the prompt template provided by Llama-3:
32
 
 
33
 
34
+ ```sh
35
+ ./llama.cpp/main -m Llama-3-13B-Instruct.Q2_K.gguf -r '<|eot_id|>' --in-prefix "\n<|start_header_id|>user<|end_header_id|>\n\n" --in-suffix "<|eot_id|><|start_header_id|>assistant<|end_header_id|>
36
+ \n\n" -p "<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nYou are a helpful, smart, kind, and efficient AI assistant. You always fulfill the user's requests to the best of your ability.<|eot_id|>\n<|start_header_id|>user<|end_header_id|>\n\nHi!<|eot_id|>\n<|start_header_id|>assistant<|end_header_id|>\n\n" -n 1024
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
37
  ```