Upload README.md
Browse files
README.md
ADDED
@@ -0,0 +1,432 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model: VAGOsolutions/SauerkrautLM-70b-v1
|
3 |
+
inference: false
|
4 |
+
language:
|
5 |
+
- de
|
6 |
+
- en
|
7 |
+
library_name: transformers
|
8 |
+
license: llama2
|
9 |
+
model_creator: VAGO solutions
|
10 |
+
model_name: SauerkrautLM 70B v1
|
11 |
+
model_type: llama
|
12 |
+
pipeline_tag: text-generation
|
13 |
+
prompt_template: "Ein Chat zwischen einem Benutzer und einem KI-Assistenten. Der KI-Assistent\
|
14 |
+
\ gibt hilfreiche, detaillierte und h\xF6fliche Antworten. \nUser: {prompt} \nAssistant:\n"
|
15 |
+
quantized_by: TheBloke
|
16 |
+
---
|
17 |
+
<!-- markdownlint-disable MD041 -->
|
18 |
+
|
19 |
+
<!-- header start -->
|
20 |
+
<!-- 200823 -->
|
21 |
+
<div style="width: auto; margin-left: auto; margin-right: auto">
|
22 |
+
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
|
23 |
+
</div>
|
24 |
+
<div style="display: flex; justify-content: space-between; width: 100%;">
|
25 |
+
<div style="display: flex; flex-direction: column; align-items: flex-start;">
|
26 |
+
<p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
|
27 |
+
</div>
|
28 |
+
<div style="display: flex; flex-direction: column; align-items: flex-end;">
|
29 |
+
<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>
|
30 |
+
</div>
|
31 |
+
</div>
|
32 |
+
<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>
|
33 |
+
<hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
|
34 |
+
<!-- header end -->
|
35 |
+
|
36 |
+
# SauerkrautLM 70B v1 - GGUF
|
37 |
+
- Model creator: [VAGO solutions](https://huggingface.co/VAGOsolutions)
|
38 |
+
- Original model: [SauerkrautLM 70B v1](https://huggingface.co/VAGOsolutions/SauerkrautLM-70b-v1)
|
39 |
+
|
40 |
+
<!-- description start -->
|
41 |
+
## Description
|
42 |
+
|
43 |
+
This repo contains GGUF format model files for [VAGO solutions's SauerkrautLM 70B v1](https://huggingface.co/VAGOsolutions/SauerkrautLM-70b-v1).
|
44 |
+
|
45 |
+
These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
|
46 |
+
|
47 |
+
<!-- description end -->
|
48 |
+
<!-- README_GGUF.md-about-gguf start -->
|
49 |
+
### About GGUF
|
50 |
+
|
51 |
+
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.
|
52 |
+
|
53 |
+
Here is an incomplate list of clients and libraries that are known to support GGUF:
|
54 |
+
|
55 |
+
* [llama.cpp](https://github.com/ggerganov/llama.cpp). The source project for GGUF. Offers a CLI and a server option.
|
56 |
+
* [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.
|
57 |
+
* [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.
|
58 |
+
* [LM Studio](https://lmstudio.ai/), an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
|
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 |
+
* [ctransformers](https://github.com/marella/ctransformers), a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
|
62 |
+
* [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
|
63 |
+
* [candle](https://github.com/huggingface/candle), a Rust ML framework with a focus on performance, including GPU support, and ease of use.
|
64 |
+
|
65 |
+
<!-- README_GGUF.md-about-gguf end -->
|
66 |
+
<!-- repositories-available start -->
|
67 |
+
## Repositories available
|
68 |
+
|
69 |
+
* [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-AWQ)
|
70 |
+
* [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GPTQ)
|
71 |
+
* [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF)
|
72 |
+
* [VAGO solutions's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/VAGOsolutions/SauerkrautLM-70b-v1)
|
73 |
+
<!-- repositories-available end -->
|
74 |
+
|
75 |
+
<!-- prompt-template start -->
|
76 |
+
## Prompt template: Sauerkraut
|
77 |
+
|
78 |
+
```
|
79 |
+
Ein Chat zwischen einem Benutzer und einem KI-Assistenten. Der KI-Assistent gibt hilfreiche, detaillierte und höfliche Antworten.
|
80 |
+
User: {prompt}
|
81 |
+
Assistant:
|
82 |
+
|
83 |
+
```
|
84 |
+
|
85 |
+
<!-- prompt-template end -->
|
86 |
+
|
87 |
+
|
88 |
+
<!-- compatibility_gguf start -->
|
89 |
+
## Compatibility
|
90 |
+
|
91 |
+
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)
|
92 |
+
|
93 |
+
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
|
94 |
+
|
95 |
+
## Explanation of quantisation methods
|
96 |
+
|
97 |
+
<details>
|
98 |
+
<summary>Click to see details</summary>
|
99 |
+
|
100 |
+
The new methods available are:
|
101 |
+
|
102 |
+
* 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)
|
103 |
+
* 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.
|
104 |
+
* 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.
|
105 |
+
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
|
106 |
+
* 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
|
107 |
+
|
108 |
+
Refer to the Provided Files table below to see what files use which methods, and how.
|
109 |
+
</details>
|
110 |
+
<!-- compatibility_gguf end -->
|
111 |
+
|
112 |
+
<!-- README_GGUF.md-provided-files start -->
|
113 |
+
## Provided files
|
114 |
+
|
115 |
+
| Name | Quant method | Bits | Size | Max RAM required | Use case |
|
116 |
+
| ---- | ---- | ---- | ---- | ---- | ----- |
|
117 |
+
| [sauerkrautlm-70b-v1.Q2_K.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q2_K.gguf) | Q2_K | 2 | 29.28 GB| 31.78 GB | smallest, significant quality loss - not recommended for most purposes |
|
118 |
+
| [sauerkrautlm-70b-v1.Q3_K_S.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q3_K_S.gguf) | Q3_K_S | 3 | 29.92 GB| 32.42 GB | very small, high quality loss |
|
119 |
+
| [sauerkrautlm-70b-v1.Q3_K_M.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q3_K_M.gguf) | Q3_K_M | 3 | 33.19 GB| 35.69 GB | very small, high quality loss |
|
120 |
+
| [sauerkrautlm-70b-v1.Q3_K_L.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q3_K_L.gguf) | Q3_K_L | 3 | 36.15 GB| 38.65 GB | small, substantial quality loss |
|
121 |
+
| [sauerkrautlm-70b-v1.Q4_0.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q4_0.gguf) | Q4_0 | 4 | 38.87 GB| 41.37 GB | legacy; small, very high quality loss - prefer using Q3_K_M |
|
122 |
+
| [sauerkrautlm-70b-v1.Q4_K_S.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q4_K_S.gguf) | Q4_K_S | 4 | 39.07 GB| 41.57 GB | small, greater quality loss |
|
123 |
+
| [sauerkrautlm-70b-v1.Q4_K_M.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q4_K_M.gguf) | Q4_K_M | 4 | 41.42 GB| 43.92 GB | medium, balanced quality - recommended |
|
124 |
+
| [sauerkrautlm-70b-v1.Q5_0.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q5_0.gguf) | Q5_0 | 5 | 47.46 GB| 49.96 GB | legacy; medium, balanced quality - prefer using Q4_K_M |
|
125 |
+
| [sauerkrautlm-70b-v1.Q5_K_S.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q5_K_S.gguf) | Q5_K_S | 5 | 47.46 GB| 49.96 GB | large, low quality loss - recommended |
|
126 |
+
| [sauerkrautlm-70b-v1.Q5_K_M.gguf](https://huggingface.co/TheBloke/SauerkrautLM-70B-v1-GGUF/blob/main/sauerkrautlm-70b-v1.Q5_K_M.gguf) | Q5_K_M | 5 | 48.75 GB| 51.25 GB | large, very low quality loss - recommended |
|
127 |
+
| sauerkrautlm-70b-v1.Q6_K.gguf | Q6_K | 6 | 56.59 GB| 59.09 GB | very large, extremely low quality loss |
|
128 |
+
| sauerkrautlm-70b-v1.Q8_0.gguf | Q8_0 | 8 | 73.29 GB| 75.79 GB | very large, extremely low quality loss - not recommended |
|
129 |
+
|
130 |
+
**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.
|
131 |
+
|
132 |
+
### Q6_K and Q8_0 files are split and require joining
|
133 |
+
|
134 |
+
**Note:** HF does not support uploading files larger than 50GB. Therefore I have uploaded the Q6_K and Q8_0 files as split files.
|
135 |
+
|
136 |
+
<details>
|
137 |
+
<summary>Click for instructions regarding Q6_K and Q8_0 files</summary>
|
138 |
+
|
139 |
+
### q6_K
|
140 |
+
Please download:
|
141 |
+
* `sauerkrautlm-70b-v1.Q6_K.gguf-split-a`
|
142 |
+
* `sauerkrautlm-70b-v1.Q6_K.gguf-split-b`
|
143 |
+
|
144 |
+
### q8_0
|
145 |
+
Please download:
|
146 |
+
* `sauerkrautlm-70b-v1.Q8_0.gguf-split-a`
|
147 |
+
* `sauerkrautlm-70b-v1.Q8_0.gguf-split-b`
|
148 |
+
|
149 |
+
To join the files, do the following:
|
150 |
+
|
151 |
+
Linux and macOS:
|
152 |
+
```
|
153 |
+
cat sauerkrautlm-70b-v1.Q6_K.gguf-split-* > sauerkrautlm-70b-v1.Q6_K.gguf && rm sauerkrautlm-70b-v1.Q6_K.gguf-split-*
|
154 |
+
cat sauerkrautlm-70b-v1.Q8_0.gguf-split-* > sauerkrautlm-70b-v1.Q8_0.gguf && rm sauerkrautlm-70b-v1.Q8_0.gguf-split-*
|
155 |
+
```
|
156 |
+
Windows command line:
|
157 |
+
```
|
158 |
+
COPY /B sauerkrautlm-70b-v1.Q6_K.gguf-split-a + sauerkrautlm-70b-v1.Q6_K.gguf-split-b sauerkrautlm-70b-v1.Q6_K.gguf
|
159 |
+
del sauerkrautlm-70b-v1.Q6_K.gguf-split-a sauerkrautlm-70b-v1.Q6_K.gguf-split-b
|
160 |
+
|
161 |
+
COPY /B sauerkrautlm-70b-v1.Q8_0.gguf-split-a + sauerkrautlm-70b-v1.Q8_0.gguf-split-b sauerkrautlm-70b-v1.Q8_0.gguf
|
162 |
+
del sauerkrautlm-70b-v1.Q8_0.gguf-split-a sauerkrautlm-70b-v1.Q8_0.gguf-split-b
|
163 |
+
```
|
164 |
+
|
165 |
+
</details>
|
166 |
+
<!-- README_GGUF.md-provided-files end -->
|
167 |
+
|
168 |
+
<!-- README_GGUF.md-how-to-download start -->
|
169 |
+
## How to download GGUF files
|
170 |
+
|
171 |
+
**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.
|
172 |
+
|
173 |
+
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
|
174 |
+
|
175 |
+
* LM Studio
|
176 |
+
* LoLLMS Web UI
|
177 |
+
* Faraday.dev
|
178 |
+
|
179 |
+
### In `text-generation-webui`
|
180 |
+
|
181 |
+
Under Download Model, you can enter the model repo: TheBloke/SauerkrautLM-70B-v1-GGUF and below it, a specific filename to download, such as: sauerkrautlm-70b-v1.Q4_K_M.gguf.
|
182 |
+
|
183 |
+
Then click Download.
|
184 |
+
|
185 |
+
### On the command line, including multiple files at once
|
186 |
+
|
187 |
+
I recommend using the `huggingface-hub` Python library:
|
188 |
+
|
189 |
+
```shell
|
190 |
+
pip3 install huggingface-hub
|
191 |
+
```
|
192 |
+
|
193 |
+
Then you can download any individual model file to the current directory, at high speed, with a command like this:
|
194 |
+
|
195 |
+
```shell
|
196 |
+
huggingface-cli download TheBloke/SauerkrautLM-70B-v1-GGUF sauerkrautlm-70b-v1.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
|
197 |
+
```
|
198 |
+
|
199 |
+
<details>
|
200 |
+
<summary>More advanced huggingface-cli download usage</summary>
|
201 |
+
|
202 |
+
You can also download multiple files at once with a pattern:
|
203 |
+
|
204 |
+
```shell
|
205 |
+
huggingface-cli download TheBloke/SauerkrautLM-70B-v1-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'
|
206 |
+
```
|
207 |
+
|
208 |
+
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).
|
209 |
+
|
210 |
+
To accelerate downloads on fast connections (1Gbit/s or higher), install `hf_transfer`:
|
211 |
+
|
212 |
+
```shell
|
213 |
+
pip3 install hf_transfer
|
214 |
+
```
|
215 |
+
|
216 |
+
And set environment variable `HF_HUB_ENABLE_HF_TRANSFER` to `1`:
|
217 |
+
|
218 |
+
```shell
|
219 |
+
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/SauerkrautLM-70B-v1-GGUF sauerkrautlm-70b-v1.Q4_K_M.gguf --local-dir . --local-dir-use-symlinks False
|
220 |
+
```
|
221 |
+
|
222 |
+
Windows Command Line users: You can set the environment variable by running `set HF_HUB_ENABLE_HF_TRANSFER=1` before the download command.
|
223 |
+
</details>
|
224 |
+
<!-- README_GGUF.md-how-to-download end -->
|
225 |
+
|
226 |
+
<!-- README_GGUF.md-how-to-run start -->
|
227 |
+
## Example `llama.cpp` command
|
228 |
+
|
229 |
+
Make sure you are using `llama.cpp` from commit [d0cee0d](https://github.com/ggerganov/llama.cpp/commit/d0cee0d36d5be95a0d9088b674dbb27354107221) or later.
|
230 |
+
|
231 |
+
```shell
|
232 |
+
./main -ngl 32 -m sauerkrautlm-70b-v1.Q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Ein Chat zwischen einem Benutzer und einem KI-Assistenten. Der KI-Assistent gibt hilfreiche, detaillierte und höfliche Antworten. \nUser: {prompt} \nAssistant:"
|
233 |
+
```
|
234 |
+
|
235 |
+
Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
|
236 |
+
|
237 |
+
Change `-c 4096` 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.
|
238 |
+
|
239 |
+
If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
|
240 |
+
|
241 |
+
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)
|
242 |
+
|
243 |
+
## How to run in `text-generation-webui`
|
244 |
+
|
245 |
+
Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
|
246 |
+
|
247 |
+
## How to run from Python code
|
248 |
+
|
249 |
+
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.
|
250 |
+
|
251 |
+
### How to load this model in Python code, using ctransformers
|
252 |
+
|
253 |
+
#### First install the package
|
254 |
+
|
255 |
+
Run one of the following commands, according to your system:
|
256 |
+
|
257 |
+
```shell
|
258 |
+
# Base ctransformers with no GPU acceleration
|
259 |
+
pip install ctransformers
|
260 |
+
# Or with CUDA GPU acceleration
|
261 |
+
pip install ctransformers[cuda]
|
262 |
+
# Or with AMD ROCm GPU acceleration (Linux only)
|
263 |
+
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
|
264 |
+
# Or with Metal GPU acceleration for macOS systems only
|
265 |
+
CT_METAL=1 pip install ctransformers --no-binary ctransformers
|
266 |
+
```
|
267 |
+
|
268 |
+
#### Simple ctransformers example code
|
269 |
+
|
270 |
+
```python
|
271 |
+
from ctransformers import AutoModelForCausalLM
|
272 |
+
|
273 |
+
# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
|
274 |
+
llm = AutoModelForCausalLM.from_pretrained("TheBloke/SauerkrautLM-70B-v1-GGUF", model_file="sauerkrautlm-70b-v1.Q4_K_M.gguf", model_type="llama", gpu_layers=50)
|
275 |
+
|
276 |
+
print(llm("AI is going to"))
|
277 |
+
```
|
278 |
+
|
279 |
+
## How to use with LangChain
|
280 |
+
|
281 |
+
Here are guides on using llama-cpp-python and ctransformers with LangChain:
|
282 |
+
|
283 |
+
* [LangChain + llama-cpp-python](https://python.langchain.com/docs/integrations/llms/llamacpp)
|
284 |
+
* [LangChain + ctransformers](https://python.langchain.com/docs/integrations/providers/ctransformers)
|
285 |
+
|
286 |
+
<!-- README_GGUF.md-how-to-run end -->
|
287 |
+
|
288 |
+
<!-- footer start -->
|
289 |
+
<!-- 200823 -->
|
290 |
+
## Discord
|
291 |
+
|
292 |
+
For further support, and discussions on these models and AI in general, join us at:
|
293 |
+
|
294 |
+
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
|
295 |
+
|
296 |
+
## Thanks, and how to contribute
|
297 |
+
|
298 |
+
Thanks to the [chirper.ai](https://chirper.ai) team!
|
299 |
+
|
300 |
+
Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
|
301 |
+
|
302 |
+
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.
|
303 |
+
|
304 |
+
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.
|
305 |
+
|
306 |
+
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
|
307 |
+
|
308 |
+
* Patreon: https://patreon.com/TheBlokeAI
|
309 |
+
* Ko-Fi: https://ko-fi.com/TheBlokeAI
|
310 |
+
|
311 |
+
**Special thanks to**: Aemon Algiz.
|
312 |
+
|
313 |
+
**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
|
314 |
+
|
315 |
+
|
316 |
+
Thank you to all my generous patrons and donaters!
|
317 |
+
|
318 |
+
And thank you again to a16z for their generous grant.
|
319 |
+
|
320 |
+
<!-- footer end -->
|
321 |
+
|
322 |
+
<!-- original-model-card start -->
|
323 |
+
# Original model card: VAGO solutions's SauerkrautLM 70B v1
|
324 |
+
|
325 |
+
|
326 |
+
![SauerkrautLM](images/SauerkrautLM.png "SauerkrautLM")
|
327 |
+
## VAGO solutions SauerkrautLM
|
328 |
+
Introducing SauerkrautLM-v1 - Your German Language Powerhouse!
|
329 |
+
|
330 |
+
We are thrilled to unveil our **very first release**, **SauerkrautLM-v1**. This remarkable creation marks a significant milestone as it is specifically **tailored for the German-speaking community**. In a landscape where German language models are scarce, we are proud to offer a solution that fills this void.
|
331 |
+
What sets SauerkrautLM-v1 apart is its versatility. Whether you are an individual looking to harness its capabilities for personal use or a business seeking to integrate it into your projects, our model is designed to accommodate all. It operates under the LLAMA 2 License, providing you with the freedom to explore its potential in both private and commercial applications.
|
332 |
+
Performance is at the heart of SauerkrautLM-v1. We put it to the **test using a customized version of MT-Bench for the German language**, and the results speak volumes. It currently stands as the most robust German Language Model on Hugging Face (based on german mt-bench results), showcasing its exceptional capabilities. Rest assured, this model is here to shine and set new standards. And the best thing is it comes in four different sizes (3B, 7B, 13B, 70B) to address your individual needs.
|
333 |
+
Our model's journey began with meticulous training using an **augmented dataset within the QLoRA approach**. This is just the beginning of our model series, promising even more innovative and powerful solutions in the future.
|
334 |
+
|
335 |
+
Join us on this exciting adventure as we redefine the possibilities of language modeling for the German-speaking world.
|
336 |
+
SauerkrautLM-v1 is here to empower your language-related endeavors like never before.
|
337 |
+
|
338 |
+
## All Models
|
339 |
+
|
340 |
+
| Model | HF | GPTQ | GGUF | AWQ |
|
341 |
+
|-------|-------|-------|-------|-------|
|
342 |
+
| SauerkrautLM-3b-v1 | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-3b-v1) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-3B-v1-GPTQ) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-3B-v1-GGUF) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-3B-v1-AWQ) |
|
343 |
+
| SauerkrautLM-7b-v1 | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-7b-v1) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7B-v1-GPTQ) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7B-v1-GGUF) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7B-v1-AWQ) |
|
344 |
+
| SauerkrautLM-7b-v1-mistral | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-7b-v1-mistral) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7b-v1-mistral-GPTQ) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7b-v1-mistral-GGUF) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-7b-v1-mistral-AWQ) |
|
345 |
+
| SauerkrautLM-13b-v1 | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-13b-v1) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-13B-v1-GPTQ) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-13B-v1-GGUF) | [Link](https://huggingface.co/TheBloke/SauerkrautLM-13B-v1-AWQ) |
|
346 |
+
| SauerkrautLM-70b-v1 | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-70b-v1) | soon | soon | soon |
|
347 |
+
|
348 |
+
## Model Details
|
349 |
+
**SauerkrautLM-70b-v1**
|
350 |
+
|
351 |
+
**Training Dataset:**
|
352 |
+
|
353 |
+
SauerkrautLM was trained with mix of German data augmentation and translated data.
|
354 |
+
We found, that only a simple translation of training data can lead to unnatural German phrasings.
|
355 |
+
Data augmentation techniques were used to grant grammatical, syntactical correctness and a more natural German wording in our training data.
|
356 |
+
|
357 |
+
**Training Procedure:**
|
358 |
+
|
359 |
+
SauerkrautLM-70b-v1 was fine-tuned using QLoRA on 2 A100 80GB with Axolotl.
|
360 |
+
|
361 |
+
- **Trained by:** SauerkrautLM-v1 trained by VAGO solutions
|
362 |
+
- **Model Type:** SauerkrautLM-v1 is an auto-regressive language model based on the transformer architecture
|
363 |
+
- **Language(s):** German, English
|
364 |
+
- **License:** [LLAMA 2 COMMUNITY LICENSE AGREEMENT](https://huggingface.co/meta-llama/Llama-2-70b/raw/main/LICENSE.txt)
|
365 |
+
- **Contact:** [Website](https://vago-solutions.de/#Kontakt) [David Golchinfar](mailto:golchinfar@vago-solutions.de)
|
366 |
+
|
367 |
+
**Prompt Template:**
|
368 |
+
|
369 |
+
```
|
370 |
+
|
371 |
+
[INST] <<SYS>>
|
372 |
+
Ein Chat zwischen einem Benutzer und einem KI-Assistenten. Der KI-Assistent gibt hilfreiche, detaillierte und höfliche Antworten.
|
373 |
+
<</SYS>>
|
374 |
+
{prompt}[/INST]
|
375 |
+
|
376 |
+
```
|
377 |
+
|
378 |
+
## Evaluation
|
379 |
+
**[MT-Bench-TrueGerman](https://huggingface.co/datasets/VAGOsolutions/MT-Bench-TrueGerman)**
|
380 |
+
|
381 |
+
![First Turn](images/FirstTurn.PNG "First Turn")
|
382 |
+
![Second Turn](images/SecondTurn.PNG "Second Turn")
|
383 |
+
![Average](images/Average.PNG "Average")
|
384 |
+
|
385 |
+
![Category Scores](images/SauerkrautLM-70b.PNG "Category Scores")
|
386 |
+
![Category Plot](images/SauerkrautLM-70b-v1.png "Category Plot")
|
387 |
+
|
388 |
+
|
389 |
+
## MT-Bench-TrueGerman Response Example
|
390 |
+
|
391 |
+
**SauerkrautLM-70b-v1**
|
392 |
+
|
393 |
+
**Prompt**
|
394 |
+
> Beschreiben Sie eine lebhafte und einzigartige Figur unter Verwendung bildhafter und kreativer Sprache. Bitte antworten Sie in weniger als zwei Absätzen.", "Überarbeiten Sie Ihre vorherige Antwort und integrieren Sie in jedem Satz eine Anspielung auf ein berühmtes literarisches Werk oder historisches Ereignis.
|
395 |
+
|
396 |
+
**Antwort SauerkrautLM-70b-v1:**
|
397 |
+
```
|
398 |
+
Sie war wie kein anderer, eine Sonnenblume inmitten von Dunkelheit. Ihr Lachen war wie ein Regenschauer nach einer langen Trockenperiode, erfrischend und erlebnisreich. Sie trat mit einem Schwung ein, ihre Haare flatterten wie Flammen, wenn sie sich zu einer Leidenschaftlichen Symphonie unter dem Mondlicht in Bewegung setzten. Ihre Augen funkelten wie Saphire, während sie die Welt mit einer Wut und Neugierde betrachteten, die man selbst in den tiefsten Ozeanen nicht finden konnte. Jeder Schritt, den sie machte, war ein Tanz, jedes Wort, das sie sprach, war eine Melodie. Sie war nicht nur eine Person, sie war ein Ereignis, ein Vorfall, eine Erfahrung, die man niemals vergessen konnte.", "Sie war wie eine moderne Helena von Troja, ihre Schönheit bezauberte alle, doch ihre Geheimnisse waren tief wie das Meer, so wie in Homer's epischer Geschichte. Ihr Lachen war wie der Klang von Glocken inmitten einer stürmischen Nacht, erinnernd an das berühmte Gedicht \"The Bells\" von Edgar Allan Poe. Mit jedem ihrer Schritte, der wie ein Tanz auf dem Wind war, erinnerte man sich an die elegante Bewegung von Cinderella in ihrer königlichen Schuhsohle. Ihre Augen funkelten wie Sterne auf der Leinwand von Van Goghs Sternennacht, während sie durch das Labyrinth des Lebens navigierte. Jedes Wort, das sie sprach, war eine Symphonie, so wie Shakespeares Sonette in den Ohren eines Liebenden klingen. Sie war nicht nur ein Mensch, sie war ein Ereignis, eine Saga, die von den Geschichtsschreibern der Zeit festgehalten wurde."
|
399 |
+
```
|
400 |
+
|
401 |
+
**Prompt**
|
402 |
+
> Sie sehen ein schönes rotes Haus auf Ihrer linken Seite und ein faszinierendes Gewächshaus auf Ihrer rechten Seite, einen attraktiven beheizten rosa Ort vor sich. Also, wo ist das Weiße Haus?", "Enthält die ursprüngliche Frage Hinweise, um den Standort des Weißen Hauses eindeutig zu bestimmen?"],"reference": ["Die Antwort ist \"Washington, DC\".", "Nein.
|
403 |
+
|
404 |
+
**Antwort SauerkrautLM-70b-v1:**
|
405 |
+
```
|
406 |
+
Das Weiße Haus ist in Washington D.C.", "Nein, die ursprüngliche Frage enthält keine Hinweise, um den Standort des Weißen Hauses eindeutig zu bestimmen."
|
407 |
+
```
|
408 |
+
|
409 |
+
**Prompt**
|
410 |
+
> Betrachten Sie einen Satelliten, der sich in einer kreisförmigen Umlaufbahn um die Erde befindet. Die Geschwindigkeit des Satelliten nimmt ab. Was wird mit dem orbitalen Radius des Satelliten und seiner Umlaufdauer passieren? Bitte rechtfertigen Sie Ihre Antwort unter Verwendung physikalischer Prinzipien.", "Was sind einige Sonderfälle oder Randfälle in Ihrer Lösung? Wie gehen Sie mit ihnen um?"] ,"reference": ["Der orbitale Radius wird zunehmen und die Umlaufdauer wird zunehmen
|
411 |
+
|
412 |
+
**Antwort SauerkrautLM-70b-v1:**
|
413 |
+
```
|
414 |
+
Wenn sich die Geschwindigkeit eines Satelliten in einer kreisförmigen Umlaufbahn um die Erde verringert, wird sich auch der Orbitalradius des Satelliten verändern. Dies kann auf Basis physikalischer Prinzipien erklärt werden.\n\n1. Erhaltung des Drehimpulses: Ein Satellit hat einen bestimmten Drehimpuls, der das Produkt aus seiner Masse, Geschwindigkeit und dem Radius entspricht. Wenn sich die Geschwindigkeit des Satelliten verringert, muss der Radius angepasst werden, um den Drehimpuls zu erhalten.\n\n2. Kepler'sche Gesetze: Johannes Kepler formulierte drei Gesetze für den Umlauf von Himmelskörpern um die Sonne, die auch auf Satelliten um die Erde angewendet werden können. Das zweite Gesetz besagt, dass die Geschwindigkeit eines Satelliten umso größer ist, je näher er an einem Fokus (in diesem Fall der Erde) ist. Wenn sich die Geschwindigkeit des Satelliten verringert, wird er sich weiter vom Fokus entfernen und somit den Orbitalradius erhöhen.\n\n3. Energieerhaltung: Die Energie eines Satelliten in einer Umlaufbahn hängt von seiner Geschwindigkeit und seinem Radius ab. Wenn die Geschwindigkeit des Satelliten abnimmt, muss die Energie aufrechterhalten werden. Dies kann nur erfolgen, indem sich der Radius erhöht, da die Geschwindigkeit bereits abgenommen hat und somit nicht mehr als Antrieb dienen kann.\n\n4. Gravitation: Die Gravitationskraft zwischen dem Satelliten und der Erde ist verantwortlich für die Umlaufbahn des Satelliten. Wenn sich der Radius des Satelliten vergrößert, verringert sich auch die Gravitationskraft zwischen dem Satelliten und der Erde. Dadurch wird die Geschwindigkeit des Satelliten weiter abgebremst, was wiederum zu weiteren Änderungen des Radius und der Umlaufdauer führen kann.\n\nBasierend auf diesen physikalischen Prinzipien lässt sich sagen, dass sich der Orbitalradius des Satelliten vergrößern wird, wenn sich seine Geschwindigkeit verringert. Dadurch wird auch die Umlaufdauer des Satelliten verändert, da die Zeit, die benötigt wird, um sich um die Erde zu bewegen, proportional zum Radius ist.", "Bei der Lösung des Problems, in dem sich die Geschwindigkeit eines Satelliten in einer kreisförmigen Umlaufbahn um die Erde verringert, gibt es einige Sonderfälle und Randfälle, die berücksichtigt werden sollten.\n\n1. Randfall: Wenn die Geschwindigkeit des Satelliten zu gering wird, kann er die Anziehungskraft der Erde nicht mehr überwinden und in die Atmosphäre eintreten. In diesem Fall würde der Satellit nicht mehr in einer Umlaufbahn bleiben, sondern in der Atmosphäre verglühen.\n\n2. Sonderfall: Wenn der Satellit in einer sehr niedrigen Umlaufbahn ist und seine Geschwindigkeit geringer als die Fluchtgeschwindigkeit für diesen Radius ist, kann es zu einer Abtragung der Geschwindigkeit durch den Widerstand in der Atmosphäre kommen. Dies kann dazu führen, dass der Satellit aus seiner Umlaufbahn abgebremst und in die Atmosphäre eintreten wird.\n\n3. Sonderfall: Wenn die Geschwindigkeit des Satelliten unter einen bestimmten Wert fällt, kann es zu einer Stabilisierung der Umlaufbahn kommen, bei der der Satellit in einer stabilen Umlaufbahn bleibt, auch wenn seine Geschwindigkeit weiter abnimmt. Dies wird als \"Hohmann-Transfer\" bezeichnet und ist ein Sonderfall, bei dem der Satellit eine Umlaufbahn um einen anderen Himmelskörper aufsuchen kann.\n\nUm mit diesen Sonderfällen und Randfällen umzugehen, ist es wichtig, die genauen Parameter des Satelliten, wie seine Masse, seine anfängliche Geschwindigkeit und seinen Orbitalradius, zu berücksichtigen. Es ist auch wichtig, die physikalischen Prinzipien, wie die Erhaltung des Drehimpulses, Keplers Gesetze, Energieerhaltung und Gravitation, zu verwenden, um die Auswirkungen der Geschwindigkeitsänderung auf den Orbitalradius und die Umlaufdauer zu analysieren."
|
415 |
+
```
|
416 |
+
|
417 |
+
|
418 |
+
## Disclaimer
|
419 |
+
Our models have been meticulously trained on extensive datasets. While we have made diligent efforts to thoroughly screen and eliminate any instances of coarse or inappropriate language from our data, we must inform users that despite our best efforts in data cleansing, the possibility of some such content slipping through cannot be entirely ruled out.
|
420 |
+
Furthermore, it is important to note that we have implemented filters within our models; however, we cannot always guarantee consistently appropriate behavior. Therefore, if you encounter any issues or come across inappropriate content, we kindly request that you inform us through the contact information provided.
|
421 |
+
Additionally, it is essential to understand that the licensing of these models does not constitute legal advice. We are not held responsible for the actions of third parties who utilize our models. These models may be employed for commercial purposes, and the original Llama2 license remains applicable and is included with the model files.
|
422 |
+
|
423 |
+
## Contact
|
424 |
+
If you are interested in customized LLMs for business applications, please get in contact with us via our website or contact us at [Dr. Daryoush Vaziri](mailto:vaziri@vago-solutions.de). We are also grateful for your feedback and suggestions.
|
425 |
+
|
426 |
+
## Collaborations
|
427 |
+
We are also keenly seeking support and investment for our startup, VAGO solutions, where we continuously advance the development of robust language models designed to address a diverse range of purposes and requirements. If the prospect of collaboratively navigating future challenges excites you, we warmly invite you to reach out to us.
|
428 |
+
|
429 |
+
## Acknowledgement
|
430 |
+
Many thanks to [TheBloke](https://huggingface.co/TheBloke) for super fast quantifying all of our models.
|
431 |
+
|
432 |
+
<!-- original-model-card end -->
|