TheBloke commited on
Commit
c001ca5
1 Parent(s): 189f1ff

Upload new k-quant GGML quantised models.

Browse files
Files changed (1) hide show
  1. README.md +69 -33
README.md CHANGED
@@ -1,13 +1,8 @@
1
  ---
2
- license: other
3
- datasets:
4
- - ehartford/wizard_vicuna_70k_unfiltered
5
- language:
6
- - en
7
- tags:
8
- - uncensored
9
  inference: false
 
10
  ---
 
11
  <!-- header start -->
12
  <div style="width: 100%;">
13
  <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
@@ -22,53 +17,89 @@ inference: false
22
  </div>
23
  <!-- header end -->
24
 
25
- # Wizard-Vicuna-13B-Uncensored GGML
26
 
27
- This is GGML format quantised 4bit and 5bit models of [Eric Hartford's 'uncensored' training of Wizard-Vicuna 13B](https://huggingface.co/ehartford/Wizard-Vicuna-13B-Uncensored).
28
 
29
- This repo is the result of quantising to 4bit and 5bit GGML for CPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp).
 
 
 
 
 
30
 
31
  ## Repositories available
32
 
33
- * [4bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-GPTQ).
34
- * [4bit and 5bit GGML models for CPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-GGML).
35
- * [float16 HF format model for GPU inference and further conversions](https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-HF).
 
 
 
 
 
 
 
 
 
 
 
36
 
37
- ## THE FILES IN MAIN BRANCH REQUIRES LATEST LLAMA.CPP (May 19th 2023 - commit 2d5db48)!
38
 
39
- llama.cpp recently made another breaking change to its quantisation methods - https://github.com/ggerganov/llama.cpp/pull/1508
40
 
41
- I have quantised the GGML files in this repo with the latest version. Therefore you will require llama.cpp compiled on May 19th or later (commit `2d5db48` or later) to use them.
42
 
43
- For files compatible with the previous version of llama.cpp, please see branch `previous_llama_ggmlv2`.
 
 
 
 
 
 
 
 
 
44
 
45
  ## Provided files
46
- | Name | Quant method | Bits | Size | RAM required | Use case |
47
  | ---- | ---- | ---- | ---- | ---- | ----- |
48
- `Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_0.bin` | q4_0 | 4bit | 8.14GB | 10.5GB | 4-bit. |
49
- `Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_1.bin` | q4_1 | 4bit | 8.95 | 11.0GB | 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models.|
50
- `Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_0.bin` | q5_0 | 5bit | 8.95GB | 11.0GB | 5-bit. Higher accuracy, higher resource usage and slower inference. |
51
- `Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_1.bin` | q5_1 | 5bit | 9.76GB | 12.25GB | 5-bit. Even higher accuracy, and higher resource usage and slower inference. |
52
- `Wizard-Vicuna-13B-Uncensored.ggmlv3.q8_0.bin` | q8_0 | 8bit | 14.6GB | 17GB | 8-bit. Almost indistinguishable from float16. Huge resource use and slow. Not recommended for normal use. |
 
 
 
 
 
 
 
 
 
 
 
 
53
 
54
  ## How to run in `llama.cpp`
55
 
56
  I use the following command line; adjust for your tastes and needs:
57
 
58
  ```
59
- ./main -t 8 -m Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: write a story about llamas ### Response:"
60
  ```
 
61
 
62
- Change `-t 8` to the number of physical CPU cores you have.
63
 
64
- ## How to run in `text-generation-webui`
65
 
66
- GGML models can be loaded into text-generation-webui by installing the llama.cpp module, then placing the ggml model file in a model folder as usual.
67
 
68
  Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md).
69
 
70
- Note: at this time text-generation-webui may not support the new May 19th llama.cpp quantisation methods for q4_0, q4_1 and q8_0 files.
71
-
72
  <!-- footer start -->
73
  ## Discord
74
 
@@ -89,19 +120,24 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
89
  * Patreon: https://patreon.com/TheBlokeAI
90
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
91
 
92
- **Patreon special mentions**: Aemon Algiz, Dmitriy Samsonov, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, Jonathan Leane, Talal Aujan, V. Lukas, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Sebastain Graf, Johann-Peter Hartman.
 
 
93
 
94
  Thank you to all my generous patrons and donaters!
 
95
  <!-- footer end -->
96
- # Original model card
 
 
97
 
98
  This is [wizard-vicuna-13b](https://huggingface.co/junelee/wizard-vicuna-13b) trained with a subset of the dataset - responses that contained alignment / moralizing were removed. The intent is to train a WizardLM that doesn't have alignment built-in, so that alignment (of any sort) can be added separately with for example with a RLHF LoRA.
99
 
100
  Shout out to the open source AI/ML community, and everyone who helped me out.
101
 
102
- Note:
103
 
104
- An uncensored model has no guardrails.
105
 
106
  You are responsible for anything you do with the model, just as you are responsible for anything you do with any dangerous object such as a knife, gun, lighter, or car.
107
 
 
1
  ---
 
 
 
 
 
 
 
2
  inference: false
3
+ license: other
4
  ---
5
+
6
  <!-- header start -->
7
  <div style="width: 100%;">
8
  <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
 
17
  </div>
18
  <!-- header end -->
19
 
20
+ # Eric Hartford's Wizard-Vicuna-13B-Uncensored GGML
21
 
22
+ These files are GGML format model files for [Eric Hartford's Wizard-Vicuna-13B-Uncensored](https://huggingface.co/ehartford/Wizard-Vicuna-13B-Uncensored).
23
 
24
+ GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp) and libraries and UIs which support this format, such as:
25
+ * [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
26
+ * [KoboldCpp](https://github.com/LostRuins/koboldcpp)
27
+ * [ParisNeo/GPT4All-UI](https://github.com/ParisNeo/gpt4all-ui)
28
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)
29
+ * [ctransformers](https://github.com/marella/ctransformers)
30
 
31
  ## Repositories available
32
 
33
+ * [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-GPTQ)
34
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/Wizard-Vicuna-13B-Uncensored-GGML)
35
+ * [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/Wizard-Vicuna-13B-Uncensored)
36
+
37
+ <!-- compatibility_ggml start -->
38
+ ## Compatibility
39
+
40
+ ### Original llama.cpp quant methods: `q4_0, q4_1, q5_0, q5_1, q8_0`
41
+
42
+ I have quantized these 'original' quantisation methods using an older version of llama.cpp so that they remain compatible with llama.cpp as of May 19th, commit `2d5db48`.
43
+
44
+ They should be compatible with all current UIs and libraries that use llama.cpp, such as those listed at the top of this README.
45
+
46
+ ### New k-quant methods: `q2_K, q3_K_S, q3_K_M, q3_K_L, q4_K_S, q4_K_M, q5_K_S, q6_K`
47
 
48
+ These new quantisation methods are only compatible with llama.cpp as of June 6th, commit `2d43387`.
49
 
50
+ They will NOT be compatible with koboldcpp, text-generation-ui, and other UIs and libraries yet. Support is expected to come over the next few days.
51
 
52
+ ## Explanation of the new k-quant methods
53
 
54
+ The new methods available are:
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
+ * GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
61
+
62
+ Refer to the Provided Files table below to see what files use which methods, and how.
63
+ <!-- compatibility_ggml end -->
64
 
65
  ## Provided files
66
+ | Name | Quant method | Bits | Size | Max RAM required | Use case |
67
  | ---- | ---- | ---- | ---- | ---- | ----- |
68
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q2_K.bin | q2_K | 2 | 5.43 GB | 7.93 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
69
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 6.87 GB | 9.37 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
70
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 6.25 GB | 8.75 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
71
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 5.59 GB | 8.09 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
72
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_0.bin | q4_0 | 4 | 7.32 GB | 9.82 GB | Original llama.cpp quant method, 4-bit. |
73
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_1.bin | q4_1 | 4 | 8.14 GB | 10.64 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
74
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 7.82 GB | 10.32 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
75
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 7.32 GB | 9.82 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
76
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_0.bin | q5_0 | 5 | 8.95 GB | 11.45 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
77
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_1.bin | q5_1 | 5 | 9.76 GB | 12.26 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
78
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 9.21 GB | 11.71 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
79
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 8.95 GB | 11.45 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
80
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q6_K.bin | q6_K | 6 | 10.68 GB | 13.18 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
81
+ | Wizard-Vicuna-13B-Uncensored.ggmlv3.q8_0.bin | q8_0 | 8 | 13.83 GB | 16.33 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
82
+
83
+
84
+ **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.
85
 
86
  ## How to run in `llama.cpp`
87
 
88
  I use the following command line; adjust for your tastes and needs:
89
 
90
  ```
91
+ ./main -t 10 -ngl 32 -m Wizard-Vicuna-13B-Uncensored.ggmlv3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: Write a story about llamas\n### Response:"
92
  ```
93
+ Change `-t 10` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`.
94
 
95
+ Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
96
 
97
+ If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
98
 
99
+ ## How to run in `text-generation-webui`
100
 
101
  Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md).
102
 
 
 
103
  <!-- footer start -->
104
  ## Discord
105
 
 
120
  * Patreon: https://patreon.com/TheBlokeAI
121
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
122
 
123
+ **Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
124
+
125
+ **Patreon special mentions**: Ajan Kanaga, Kalila, Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.
126
 
127
  Thank you to all my generous patrons and donaters!
128
+
129
  <!-- footer end -->
130
+
131
+ # Original model card: Eric Hartford's Wizard-Vicuna-13B-Uncensored
132
+
133
 
134
  This is [wizard-vicuna-13b](https://huggingface.co/junelee/wizard-vicuna-13b) trained with a subset of the dataset - responses that contained alignment / moralizing were removed. The intent is to train a WizardLM that doesn't have alignment built-in, so that alignment (of any sort) can be added separately with for example with a RLHF LoRA.
135
 
136
  Shout out to the open source AI/ML community, and everyone who helped me out.
137
 
138
+ Note:
139
 
140
+ An uncensored model has no guardrails.
141
 
142
  You are responsible for anything you do with the model, just as you are responsible for anything you do with any dangerous object such as a knife, gun, lighter, or car.
143