TheBloke commited on
Commit
315eefb
1 Parent(s): d5c7620

Initial GGML model commit

Browse files
Files changed (1) hide show
  1. README.md +69 -30
README.md CHANGED
@@ -1,11 +1,8 @@
1
  ---
2
- license: other
3
- datasets:
4
- - ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered
5
- tags:
6
- - uncensored
7
  inference: false
 
8
  ---
 
9
  <!-- header start -->
10
  <div style="width: 100%;">
11
  <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
@@ -20,44 +17,82 @@ inference: false
20
  </div>
21
  <!-- header end -->
22
 
23
- # WizardLM 30B uncensored:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24
 
25
- These files are GGML format model files for [Eric Hartford's 'uncensored' 30B version of WizardLM](https://huggingface.co/ehartford/WizardLM-30B-Uncensored).
 
26
 
27
- GGML files are for CPU inference using [llama.cpp](https://github.com/ggerganov/llama.cpp).
28
 
29
- ## Other repositories available
30
 
31
- * [4bit GPTQ model for GPU inference](https://huggingface.co/TheBloke/WizardLM-30B-Uncensored-GPTQ)
32
- * [4-bit, 5-bit and 8-bit GGML models for CPU (+CUDA) inference](https://huggingface.co/TheBloke/WizardLM-30B-Uncensored-GGML)
33
- * [Eric's unquantised model in fp16 HF format](https://huggingface.co/ehartford/WizardLM-30B-Uncensored)
34
 
35
- ## THE FILES IN MAIN BRANCH REQUIRES LATEST LLAMA.CPP (May 19th 2023 - commit 2d5db48)!
36
 
37
- llama.cpp recently made another breaking change to its quantisation methods - https://github.com/ggerganov/llama.cpp/pull/1508
38
 
39
- 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.
 
 
 
 
 
 
 
 
 
 
 
 
 
40
 
41
  ## Provided files
42
- | Name | Quant method | Bits | Size | RAM required | Use case |
43
  | ---- | ---- | ---- | ---- | ---- | ----- |
44
- `WizardLM-30B-Uncensored.ggmlv3.q4_0.bin` | q4_0 | 4bit | 18.3GB | 20GB | 4-bit. |
45
- `WizardLM-30B-Uncensored.ggmlv3.q4_1.bin` | q4_1 | 4bit | 20.3GB | 23GB | 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
46
- `WizardLM-30B-Uncensored.ggmlv3.q5_0.bin` | q5_0 | 5bit | 22.4GB | 25GB | 5-bit. Higher accuracy, higher resource usage, slower inference. |
47
- `WizardLM-30B-Uncensored.ggmlv3.q5_1.bin` | q5_1 | 5bit | 24.4GB | 27GB | 5-bit. Even higher accuracy and resource usage, and slower inference. |
48
- `WizardLM-30B-Uncensored.ggmlv3.q8_0.bin` | q8_0 | 8bit | 34.6GB | 38GB | 8-bit. Almost indistinguishable from float16. Huge resource use and slow. Not recommended for normal use.|
 
 
 
 
 
 
 
 
 
 
 
 
49
 
50
  ## How to run in `llama.cpp`
51
 
52
  I use the following command line; adjust for your tastes and needs:
53
 
54
  ```
55
- ./main -t 12 -m WizardLM-30B-Uncensored.ggmlv3.q4_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Below is an instruction that describes a task. Write a response that appropriately completes the request.
56
- ### Instruction:
57
- Write a story about llamas
58
- ### Response:"
59
  ```
60
- Change `-t 12` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`.
 
 
61
 
62
  If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
63
 
@@ -85,18 +120,22 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
85
  * Patreon: https://patreon.com/TheBlokeAI
86
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
87
 
88
- **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.
 
 
89
 
90
  Thank you to all my generous patrons and donaters!
 
91
  <!-- footer end -->
92
- # WizardLM-30B-Uncensored original model card
 
93
 
94
  This is WizardLM 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.
95
 
96
  Shout out to the open source AI/ML community, and everyone who helped me out.
97
 
98
- Note:
99
- An uncensored model has no guardrails.
100
  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.
101
  Publishing anything this model generates is the same as publishing it yourself.
102
  You are responsible for the content you publish, and you cannot blame the model any more than you can blame the knife, gun, lighter, or car for what you do with it.
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 'uncensored' WizardLM 30B GGML
21
+
22
+ These files are GGML format model files for [Eric Hartford's 'uncensored' WizardLM 30B](https://huggingface.co/ehartford/WizardLM-30B-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/WizardLM-30B-Uncensored-GPTQ)
34
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/WizardLM-30B-Uncensored-GGML)
35
+ * [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/WizardLM-30B-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
+ | WizardLM-30B-Uncensored.ggmlv3.q4_0.bin | q4_0 | 4 | 18.30 GB | 20.80 GB | Original llama.cpp quant method, 4-bit. |
69
+ | WizardLM-30B-Uncensored.ggmlv3.q4_1.bin | q4_1 | 4 | 20.33 GB | 22.83 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. |
70
+ | WizardLM-30B-Uncensored.ggmlv3.q5_0.bin | q5_0 | 5 | 22.37 GB | 24.87 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
71
+ | WizardLM-30B-Uncensored.ggmlv3.q5_1.bin | q5_1 | 5 | 24.40 GB | 26.90 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
72
+ | WizardLM-30B-Uncensored.ggmlv3.q8_0.bin | q8_0 | 8 | 34.56 GB | 37.06 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
73
+ | wizardlm-30b-uncensored.ggmlv3.q2_K.bin | q2_K | 2 | 13.60 GB | 16.10 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. |
74
+ | wizardlm-30b-uncensored.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 17.20 GB | 19.70 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 |
75
+ | wizardlm-30b-uncensored.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 15.64 GB | 18.14 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 |
76
+ | wizardlm-30b-uncensored.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 13.98 GB | 16.48 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
77
+ | wizardlm-30b-uncensored.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 19.57 GB | 22.07 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 |
78
+ | wizardlm-30b-uncensored.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 18.30 GB | 20.80 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
79
+ | wizardlm-30b-uncensored.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 23.02 GB | 25.52 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 |
80
+ | wizardlm-30b-uncensored.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 22.37 GB | 24.87 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
81
+ | wizardlm-30b-uncensored.ggmlv3.q6_K.bin | q6_K | 6 | 26.69 GB | 29.19 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
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 wizardlm-30b-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
 
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 'uncensored' WizardLM 30B
132
 
133
  This is WizardLM 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.
134
 
135
  Shout out to the open source AI/ML community, and everyone who helped me out.
136
 
137
+ Note:
138
+ An uncensored model has no guardrails.
139
  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.
140
  Publishing anything this model generates is the same as publishing it yourself.
141
  You are responsible for the content you publish, and you cannot blame the model any more than you can blame the knife, gun, lighter, or car for what you do with it.