Transformers
llama
text-generation-inference
TheBloke commited on
Commit
c018fb0
β€’
1 Parent(s): 52e0d36

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +118 -27
README.md CHANGED
@@ -1,6 +1,6 @@
1
  ---
2
  inference: false
3
- license: other
4
  model_creator: WizardLM
5
  model_link: https://huggingface.co/WizardLM/WizardLM-13B-V1.2
6
  model_name: WizardLM 13B V1.2
@@ -9,17 +9,20 @@ quantized_by: TheBloke
9
  ---
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;">
 
14
  </div>
15
  <div style="display: flex; justify-content: space-between; width: 100%;">
16
  <div style="display: flex; flex-direction: column; align-items: flex-start;">
17
- <p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
18
  </div>
19
  <div style="display: flex; flex-direction: column; align-items: flex-end;">
20
- <p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
21
  </div>
22
  </div>
 
 
23
  <!-- header end -->
24
 
25
  # WizardLM 13B V1.2 - GGML
@@ -30,6 +33,13 @@ quantized_by: TheBloke
30
 
31
  This repo contains GGML format model files for [WizardLM's WizardLM 13B V1.2](https://huggingface.co/WizardLM/WizardLM-13B-V1.2).
32
 
 
 
 
 
 
 
 
33
  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:
34
  * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most popular web UI. Supports NVidia CUDA GPU acceleration.
35
  * [KoboldCpp](https://github.com/LostRuins/koboldcpp), a powerful GGML web UI with GPU acceleration on all platforms (CUDA and OpenCL). Especially good for story telling.
@@ -41,30 +51,27 @@ GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/gger
41
  ## Repositories available
42
 
43
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GPTQ)
44
- * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML)
 
45
  * [WizardLM's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/WizardLM/WizardLM-13B-V1.2)
46
 
47
  ## Prompt template: Vicuna
48
 
49
  ```
50
- A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions.
51
 
52
- USER: {prompt}
53
- ASSISTANT:
54
  ```
55
 
56
  <!-- compatibility_ggml start -->
57
  ## Compatibility
58
 
59
- ### Original llama.cpp quant methods: `q4_0, q4_1, q5_0, q5_1, q8_0`
60
 
61
- These are guaranteed to be compatible with any UIs, tools and libraries released since late May. They may be phased out soon, as they are largely superseded by the new k-quant methods.
62
 
63
- ### 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`
64
 
65
- These new quantisation methods are compatible with llama.cpp as of June 6th, commit `2d43387`.
66
-
67
- They are now also compatible with recent releases of text-generation-webui, KoboldCpp, llama-cpp-python, ctransformers, rustformers and most others. For compatibility with other tools and libraries, please check their documentation.
68
 
69
  ## Explanation of the new k-quant methods
70
  <details>
@@ -83,20 +90,21 @@ Refer to the Provided Files table below to see what files use which methods, and
83
  <!-- compatibility_ggml end -->
84
 
85
  ## Provided files
 
86
  | Name | Quant method | Bits | Size | Max RAM required | Use case |
87
  | ---- | ---- | ---- | ---- | ---- | ----- |
88
  | [wizardlm-13b-v1.2.ggmlv3.q2_K.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q2_K.bin) | q2_K | 2 | 5.51 GB| 8.01 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. |
89
- | [wizardlm-13b-v1.2.ggmlv3.q3_K_L.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_L.bin) | q3_K_L | 3 | 6.93 GB| 9.43 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 |
90
- | [wizardlm-13b-v1.2.ggmlv3.q3_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_M.bin) | q3_K_M | 3 | 6.31 GB| 8.81 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 |
91
  | [wizardlm-13b-v1.2.ggmlv3.q3_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_S.bin) | q3_K_S | 3 | 5.66 GB| 8.16 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
 
 
92
  | [wizardlm-13b-v1.2.ggmlv3.q4_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_0.bin) | q4_0 | 4 | 7.32 GB| 9.82 GB | Original quant method, 4-bit. |
93
- | [wizardlm-13b-v1.2.ggmlv3.q4_1.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_1.bin) | q4_1 | 4 | 8.14 GB| 10.64 GB | Original quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
94
- | [wizardlm-13b-v1.2.ggmlv3.q4_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_K_M.bin) | q4_K_M | 4 | 7.87 GB| 10.37 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 |
95
  | [wizardlm-13b-v1.2.ggmlv3.q4_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_K_S.bin) | q4_K_S | 4 | 7.37 GB| 9.87 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
 
 
96
  | [wizardlm-13b-v1.2.ggmlv3.q5_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_0.bin) | q5_0 | 5 | 8.95 GB| 11.45 GB | Original quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
97
- | [wizardlm-13b-v1.2.ggmlv3.q5_1.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_1.bin) | q5_1 | 5 | 9.76 GB| 12.26 GB | Original quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
98
- | [wizardlm-13b-v1.2.ggmlv3.q5_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_K_M.bin) | q5_K_M | 5 | 9.23 GB| 11.73 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 |
99
  | [wizardlm-13b-v1.2.ggmlv3.q5_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_K_S.bin) | q5_K_S | 5 | 8.97 GB| 11.47 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
 
 
100
  | [wizardlm-13b-v1.2.ggmlv3.q6_K.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q6_K.bin) | q6_K | 6 | 10.68 GB| 13.18 GB | New k-quant method. Uses GGML_TYPE_Q8_K for all tensors - 6-bit quantization |
101
  | [wizardlm-13b-v1.2.ggmlv3.q8_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q8_0.bin) | q8_0 | 8 | 13.83 GB| 16.33 GB | Original quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
102
 
@@ -104,22 +112,29 @@ Refer to the Provided Files table below to see what files use which methods, and
104
 
105
  ## How to run in `llama.cpp`
106
 
107
- I use the following command line; adjust for your tastes and needs:
 
 
108
 
109
  ```
110
- ./main -t 10 -ngl 32 -m wizardlm-13b-v1.2.ggmlv3.q4_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: Write a story about llamas\n### Response:"
111
  ```
112
  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`.
113
 
114
  Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
115
 
 
 
116
  If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
117
 
 
 
118
  ## How to run in `text-generation-webui`
119
 
120
- 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).
121
 
122
  <!-- footer start -->
 
123
  ## Discord
124
 
125
  For further support, and discussions on these models and AI in general, join us at:
@@ -139,18 +154,70 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
139
  * Patreon: https://patreon.com/TheBlokeAI
140
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
141
 
142
- **Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
143
 
144
- **Patreon special mentions**: Slarti, Chadd, John Detwiler, Pieter, zynix, K, Mano Prime, ReadyPlayerEmma, Ai Maven, Leonard Tan, Edmond Seymore, Joseph William Delisle, Luke @flexchar, Fred von Graf, Viktor Bowallius, Rishabh Srivastava, Nikolai Manek, Matthew Berman, Johann-Peter Hartmann, ya boyyy, Greatston Gnanesh, Femi Adebogun, Talal Aujan, Jonathan Leane, terasurfer, David Flickinger, William Sang, Ajan Kanaga, Vadim, Artur Olbinski, Raven Klaugh, Michael Levine, Oscar Rangel, Randy H, Cory Kujawski, RoA, Dave, Alex, Alexandros Triantafyllidis, Fen Risland, Eugene Pentland, vamX, Elle, Nathan LeClaire, Khalefa Al-Ahmad, Rainer Wilmers, subjectnull, Junyu Yang, Daniel P. Andersen, SuperWojo, LangChain4j, Mandus, Kalila, Illia Dulskyi, Trenton Dambrowitz, Asp the Wyvern, Derek Yates, Jeffrey Morgan, Deep Realms, Imad Khwaja, Pyrater, Preetika Verma, biorpg, Gabriel Tamborski, Stephen Murray, Spiking Neurons AB, Iucharbius, Chris Smitley, Willem Michiel, Luke Pendergrass, Sebastain Graf, senxiiz, Will Dee, Space Cruiser, Karl Bernard, Clay Pascal, Lone Striker, transmissions 11, webtim, WelcomeToTheClub, Sam, theTransient, Pierre Kircher, chris gileta, John Villwock, Sean Connelly, Willian Hasse
145
 
146
 
147
  Thank you to all my generous patrons and donaters!
148
 
 
 
149
  <!-- footer end -->
150
 
151
  # Original model card: WizardLM's WizardLM 13B V1.2
152
 
153
- This is the **Full-Weight** of WizardLM-13B V1.2 model.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
154
 
155
  **Repository**: https://github.com/nlpxucan/WizardLM
156
 
@@ -159,3 +226,27 @@ This is the **Full-Weight** of WizardLM-13B V1.2 model.
159
 
160
  - πŸ”₯πŸ”₯πŸ”₯ [7/25/2023] We released **WizardLM V1.2** models. The **WizardLM-13B-V1.2** is here ([Demo_13B-V1.2](https://b7a19878988c8c73.gradio.app), [Demo_13B-V1.2_bak-1](https://d0a37a76e0ac4b52.gradio.app/), [Full Model Weight](https://huggingface.co/WizardLM/WizardLM-13B-V1.2)). Please checkout the [paper](https://arxiv.org/abs/2304.12244).
161
  - πŸ”₯πŸ”₯πŸ”₯ [7/25/2023] The **WizardLM-13B-V1.2** achieves **7.06** on [MT-Bench Leaderboard](https://chat.lmsys.org/?leaderboard), **89.17%** on [AlpacaEval Leaderboard](https://tatsu-lab.github.io/alpaca_eval/), and **101.4%** on [WizardLM Eval](https://github.com/nlpxucan/WizardLM/blob/main/WizardLM/data/WizardLM_testset.jsonl). (Note: MT-Bench and AlpacaEval are all self-test, will push update and request review. All tests are completed under their official settings.)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  inference: false
3
+ license: llama2
4
  model_creator: WizardLM
5
  model_link: https://huggingface.co/WizardLM/WizardLM-13B-V1.2
6
  model_name: WizardLM 13B V1.2
 
9
  ---
10
 
11
  <!-- header start -->
12
+ <!-- 200823 -->
13
+ <div style="width: auto; margin-left: auto; margin-right: auto">
14
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
15
  </div>
16
  <div style="display: flex; justify-content: space-between; width: 100%;">
17
  <div style="display: flex; flex-direction: column; align-items: flex-start;">
18
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
19
  </div>
20
  <div style="display: flex; flex-direction: column; align-items: flex-end;">
21
+ <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>
22
  </div>
23
  </div>
24
+ <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>
25
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
26
  <!-- header end -->
27
 
28
  # WizardLM 13B V1.2 - GGML
 
33
 
34
  This repo contains GGML format model files for [WizardLM's WizardLM 13B V1.2](https://huggingface.co/WizardLM/WizardLM-13B-V1.2).
35
 
36
+ ### Important note regarding GGML files.
37
+
38
+ The GGML format has now been superseded by GGUF. As of August 21st 2023, [llama.cpp](https://github.com/ggerganov/llama.cpp) no longer supports GGML models. Third party clients and libraries are expected to still support it for a time, but many may also drop support.
39
+
40
+ Please use the GGUF models instead.
41
+ ### About GGML
42
+
43
  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:
44
  * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most popular web UI. Supports NVidia CUDA GPU acceleration.
45
  * [KoboldCpp](https://github.com/LostRuins/koboldcpp), a powerful GGML web UI with GPU acceleration on all platforms (CUDA and OpenCL). Especially good for story telling.
 
51
  ## Repositories available
52
 
53
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GPTQ)
54
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGUF)
55
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML)
56
  * [WizardLM's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/WizardLM/WizardLM-13B-V1.2)
57
 
58
  ## Prompt template: Vicuna
59
 
60
  ```
61
+ A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: {prompt} ASSISTANT:
62
 
 
 
63
  ```
64
 
65
  <!-- compatibility_ggml start -->
66
  ## Compatibility
67
 
68
+ These quantised GGML files are compatible with llama.cpp between June 6th (commit `2d43387`) and August 21st 2023.
69
 
70
+ For support with latest llama.cpp, please use GGUF files instead.
71
 
72
+ The final llama.cpp commit with support for GGML was: [dadbed99e65252d79f81101a392d0d6497b86caa](https://github.com/ggerganov/llama.cpp/commit/dadbed99e65252d79f81101a392d0d6497b86caa)
73
 
74
+ As of August 23rd 2023 they are still compatible with all UIs, libraries and utilities which use GGML. This may change in the future.
 
 
75
 
76
  ## Explanation of the new k-quant methods
77
  <details>
 
90
  <!-- compatibility_ggml end -->
91
 
92
  ## Provided files
93
+
94
  | Name | Quant method | Bits | Size | Max RAM required | Use case |
95
  | ---- | ---- | ---- | ---- | ---- | ----- |
96
  | [wizardlm-13b-v1.2.ggmlv3.q2_K.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q2_K.bin) | q2_K | 2 | 5.51 GB| 8.01 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. |
 
 
97
  | [wizardlm-13b-v1.2.ggmlv3.q3_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_S.bin) | q3_K_S | 3 | 5.66 GB| 8.16 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
98
+ | [wizardlm-13b-v1.2.ggmlv3.q3_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_M.bin) | q3_K_M | 3 | 6.31 GB| 8.81 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 |
99
+ | [wizardlm-13b-v1.2.ggmlv3.q3_K_L.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q3_K_L.bin) | q3_K_L | 3 | 6.93 GB| 9.43 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 |
100
  | [wizardlm-13b-v1.2.ggmlv3.q4_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_0.bin) | q4_0 | 4 | 7.32 GB| 9.82 GB | Original quant method, 4-bit. |
 
 
101
  | [wizardlm-13b-v1.2.ggmlv3.q4_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_K_S.bin) | q4_K_S | 4 | 7.37 GB| 9.87 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
102
+ | [wizardlm-13b-v1.2.ggmlv3.q4_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_K_M.bin) | q4_K_M | 4 | 7.87 GB| 10.37 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 |
103
+ | [wizardlm-13b-v1.2.ggmlv3.q4_1.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q4_1.bin) | q4_1 | 4 | 8.14 GB| 10.64 GB | Original quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
104
  | [wizardlm-13b-v1.2.ggmlv3.q5_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_0.bin) | q5_0 | 5 | 8.95 GB| 11.45 GB | Original quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
 
 
105
  | [wizardlm-13b-v1.2.ggmlv3.q5_K_S.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_K_S.bin) | q5_K_S | 5 | 8.97 GB| 11.47 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
106
+ | [wizardlm-13b-v1.2.ggmlv3.q5_K_M.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_K_M.bin) | q5_K_M | 5 | 9.23 GB| 11.73 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 |
107
+ | [wizardlm-13b-v1.2.ggmlv3.q5_1.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q5_1.bin) | q5_1 | 5 | 9.76 GB| 12.26 GB | Original quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
108
  | [wizardlm-13b-v1.2.ggmlv3.q6_K.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q6_K.bin) | q6_K | 6 | 10.68 GB| 13.18 GB | New k-quant method. Uses GGML_TYPE_Q8_K for all tensors - 6-bit quantization |
109
  | [wizardlm-13b-v1.2.ggmlv3.q8_0.bin](https://huggingface.co/TheBloke/WizardLM-13B-V1.2-GGML/blob/main/wizardlm-13b-v1.2.ggmlv3.q8_0.bin) | q8_0 | 8 | 13.83 GB| 16.33 GB | Original quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
110
 
 
112
 
113
  ## How to run in `llama.cpp`
114
 
115
+ Make sure you are using `llama.cpp` from commit [dadbed99e65252d79f81101a392d0d6497b86caa](https://github.com/ggerganov/llama.cpp/commit/dadbed99e65252d79f81101a392d0d6497b86caa) or earlier.
116
+
117
+ For compatibility with latest llama.cpp, please use GGUF files instead.
118
 
119
  ```
120
+ ./main -t 10 -ngl 32 -m wizardlm-13b-v1.2.ggmlv3.q4_K_M.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: Write a story about llamas ASSISTANT:"
121
  ```
122
  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`.
123
 
124
  Change `-ngl 32` to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
125
 
126
+ Change `-c 2048` to the desired sequence length for this model. For example, `-c 4096` for a Llama 2 model. For models that use RoPE, add `--rope-freq-base 10000 --rope-freq-scale 0.5` for doubled context, or `--rope-freq-base 10000 --rope-freq-scale 0.25` for 4x context.
127
+
128
  If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`
129
 
130
+ 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)
131
+
132
  ## How to run in `text-generation-webui`
133
 
134
+ Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
135
 
136
  <!-- footer start -->
137
+ <!-- 200823 -->
138
  ## Discord
139
 
140
  For further support, and discussions on these models and AI in general, join us at:
 
154
  * Patreon: https://patreon.com/TheBlokeAI
155
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
156
 
157
+ **Special thanks to**: Aemon Algiz.
158
 
159
+ **Patreon special mentions**: Russ Johnson, J, alfie_i, Alex, NimbleBox.ai, Chadd, Mandus, Nikolai Manek, Ken Nordquist, ya boyyy, Illia Dulskyi, Viktor Bowallius, vamX, Iucharbius, zynix, Magnesian, Clay Pascal, Pierre Kircher, Enrico Ros, Tony Hughes, Elle, Andrey, knownsqashed, Deep Realms, Jerry Meng, Lone Striker, Derek Yates, Pyrater, Mesiah Bishop, James Bentley, Femi Adebogun, Brandon Frisco, SuperWojo, Alps Aficionado, Michael Dempsey, Vitor Caleffi, Will Dee, Edmond Seymore, usrbinkat, LangChain4j, Kacper WikieΕ‚, Luke Pendergrass, John Detwiler, theTransient, Nathan LeClaire, Tiffany J. Kim, biorpg, Eugene Pentland, Stanislav Ovsiannikov, Fred von Graf, terasurfer, Kalila, Dan Guido, Nitin Borwankar, 阿明, Ai Maven, John Villwock, Gabriel Puliatti, Stephen Murray, Asp the Wyvern, danny, Chris Smitley, ReadyPlayerEmma, S_X, Daniel P. Andersen, Olakabola, Jeffrey Morgan, Imad Khwaja, Caitlyn Gatomon, webtim, Alicia Loh, Trenton Dambrowitz, Swaroop Kallakuri, Erik BjΓ€reholt, Leonard Tan, Spiking Neurons AB, Luke @flexchar, Ajan Kanaga, Thomas Belote, Deo Leter, RoA, Willem Michiel, transmissions 11, subjectnull, Matthew Berman, Joseph William Delisle, David Ziegler, Michael Davis, Johann-Peter Hartmann, Talal Aujan, senxiiz, Artur Olbinski, Rainer Wilmers, Spencer Kim, Fen Risland, Cap'n Zoog, Rishabh Srivastava, Michael Levine, Geoffrey Montalvo, Sean Connelly, Alexandros Triantafyllidis, Pieter, Gabriel Tamborski, Sam, Subspace Studios, Junyu Yang, Pedro Madruga, Vadim, Cory Kujawski, K, Raven Klaugh, Randy H, Mano Prime, Sebastain Graf, Space Cruiser
160
 
161
 
162
  Thank you to all my generous patrons and donaters!
163
 
164
+ And thank you again to a16z for their generous grant.
165
+
166
  <!-- footer end -->
167
 
168
  # Original model card: WizardLM's WizardLM 13B V1.2
169
 
170
+ This is the **Full-Weight** of WizardLM-13B V1.2 model, this model is trained from **Llama-2 13b**.
171
+
172
+ ## WizardLM: Empowering Large Pre-Trained Language Models to Follow Complex Instructions
173
+
174
+
175
+
176
+ <p align="center">
177
+ πŸ€— <a href="https://huggingface.co/WizardLM" target="_blank">HF Repo</a> β€’πŸ± <a href="https://github.com/nlpxucan/WizardLM" target="_blank">Github Repo</a> β€’ 🐦 <a href="https://twitter.com/WizardLM_AI" target="_blank">Twitter</a> β€’ πŸ“ƒ <a href="https://arxiv.org/abs/2304.12244" target="_blank">[WizardLM]</a> β€’ πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> β€’ πŸ“ƒ <a href="https://arxiv.org/abs/2308.09583" target="_blank">[WizardMath]</a> <br>
178
+ </p>
179
+ <p align="center">
180
+ πŸ‘‹ Join our <a href="https://discord.gg/VZjjHtWrKs" target="_blank">Discord</a>
181
+ </p>
182
+
183
+ ## News
184
+
185
+ - πŸ”₯πŸ”₯πŸ”₯[2023/08/26] We released **WizardCoder-Python-34B-V1.0** , which achieves the **73.2 pass@1** and surpasses **GPT4 (2023/03/15)**, **ChatGPT-3.5**, and **Claude2** on the [HumanEval Benchmarks](https://github.com/openai/human-eval). For more details, please refer to [WizardCoder](https://github.com/nlpxucan/WizardLM/tree/main/WizardCoder).
186
+ - [2023/06/16] We released **WizardCoder-15B-V1.0** , which surpasses **Claude-Plus (+6.8)**, **Bard (+15.3)** and **InstructCodeT5+ (+22.3)** on the [HumanEval Benchmarks](https://github.com/openai/human-eval). For more details, please refer to [WizardCoder](https://github.com/nlpxucan/WizardLM/tree/main/WizardCoder).
187
+
188
+ | Model | Checkpoint | Paper | HumanEval | MBPP | Demo | License |
189
+ | ----- |------| ---- |------|-------| ----- | ----- |
190
+ | WizardCoder-Python-34B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-Python-34B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 73.2 | 61.2 | [Demo](http://47.103.63.15:50085/) | <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama2</a> |
191
+ | WizardCoder-15B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-15B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 59.8 |50.6 | -- | <a href="https://huggingface.co/spaces/bigcode/bigcode-model-license-agreement" target="_blank">OpenRAIL-M</a> |
192
+ | WizardCoder-Python-13B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-Python-13B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 64.0 | 55.6 | -- | <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama2</a> |
193
+ | WizardCoder-Python-7B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-Python-7B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 55.5 | 51.6 | [Demo](http://47.103.63.15:50088/) | <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama2</a> |
194
+ | WizardCoder-3B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-3B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 34.8 |37.4 | -- | <a href="https://huggingface.co/spaces/bigcode/bigcode-model-license-agreement" target="_blank">OpenRAIL-M</a> |
195
+ | WizardCoder-1B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardCoder-1B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2306.08568" target="_blank">[WizardCoder]</a> | 23.8 |28.6 | -- | <a href="https://huggingface.co/spaces/bigcode/bigcode-model-license-agreement" target="_blank">OpenRAIL-M</a> |
196
+
197
+
198
+ - πŸ”₯ [08/11/2023] We release **WizardMath** Models.
199
+ - πŸ”₯ Our **WizardMath-70B-V1.0** model slightly outperforms some closed-source LLMs on the GSM8K, including **ChatGPT 3.5**, **Claude Instant 1** and **PaLM 2 540B**.
200
+ - πŸ”₯ Our **WizardMath-70B-V1.0** model achieves **81.6 pass@1** on the [GSM8k Benchmarks](https://github.com/openai/grade-school-math), which is **24.8** points higher than the SOTA open-source LLM.
201
+ - πŸ”₯ Our **WizardMath-70B-V1.0** model achieves **22.7 pass@1** on the [MATH Benchmarks](https://github.com/hendrycks/math), which is **9.2** points higher than the SOTA open-source LLM.
202
+
203
+
204
+ | Model | Checkpoint | Paper | GSM8k | MATH |Online Demo| License|
205
+ | ----- |------| ---- |------|-------| ----- | ----- |
206
+ | WizardMath-70B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardMath-70B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2308.09583" target="_blank">[WizardMath]</a>| **81.6** | **22.7** |[Demo](http://47.103.63.15:50083/)| <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama 2 </a> |
207
+ | WizardMath-13B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardMath-13B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2308.09583" target="_blank">[WizardMath]</a>| **63.9** | **14.0** |[Demo](http://47.103.63.15:50082/)| <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama 2 </a> |
208
+ | WizardMath-7B-V1.0 | πŸ€— <a href="https://huggingface.co/WizardLM/WizardMath-7B-V1.0" target="_blank">HF Link</a> | πŸ“ƒ <a href="https://arxiv.org/abs/2308.09583" target="_blank">[WizardMath]</a>| **54.9** | **10.7** | [Demo](http://47.103.63.15:50080/)| <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama 2 </a>|
209
+
210
+
211
+ <font size=4>
212
+
213
+ | <sup>Model</sup> | <sup>Checkpoint</sup> | <sup>Paper</sup> |<sup>MT-Bench</sup> | <sup>AlpacaEval</sup> | <sup>WizardEval</sup> | <sup>HumanEval</sup> | <sup>License</sup>|
214
+ | ----- |------| ---- |------|-------| ----- | ----- | ----- |
215
+ | <sup>WizardLM-13B-V1.2</sup> | <sup>πŸ€— <a href="https://huggingface.co/WizardLM/WizardLM-13B-V1.2" target="_blank">HF Link</a> </sup>| | <sup>7.06</sup> | <sup>89.17%</sup> | <sup>101.4% </sup>|<sup>36.6 pass@1</sup>|<sup> <a href="https://ai.meta.com/resources/models-and-libraries/llama-downloads/" target="_blank">Llama 2 License </a></sup> |
216
+ | <sup>WizardLM-13B-V1.1</sup> |<sup> πŸ€— <a href="https://huggingface.co/WizardLM/WizardLM-13B-V1.1" target="_blank">HF Link</a> </sup> | | <sup>6.76</sup> |<sup>86.32%</sup> | <sup>99.3% </sup> |<sup>25.0 pass@1</sup>| <sup>Non-commercial</sup>|
217
+ | <sup>WizardLM-30B-V1.0</sup> | <sup>πŸ€— <a href="https://huggingface.co/WizardLM/WizardLM-30B-V1.0" target="_blank">HF Link</a></sup> | | <sup>7.01</sup> | | <sup>97.8% </sup> | <sup>37.8 pass@1</sup>| <sup>Non-commercial</sup> |
218
+ | <sup>WizardLM-13B-V1.0</sup> | <sup>πŸ€— <a href="https://huggingface.co/WizardLM/WizardLM-13B-V1.0" target="_blank">HF Link</a> </sup> | | <sup>6.35</sup> | <sup>75.31%</sup> | <sup>89.1% </sup> |<sup> 24.0 pass@1 </sup> | <sup>Non-commercial</sup>|
219
+ | <sup>WizardLM-7B-V1.0 </sup>| <sup>πŸ€— <a href="https://huggingface.co/WizardLM/WizardLM-7B-V1.0" target="_blank">HF Link</a> </sup> |<sup> πŸ“ƒ <a href="https://arxiv.org/abs/2304.12244" target="_blank">[WizardLM]</a> </sup>| | | <sup>78.0% </sup> |<sup>19.1 pass@1 </sup>|<sup> Non-commercial</sup>|
220
+ </font>
221
 
222
  **Repository**: https://github.com/nlpxucan/WizardLM
223
 
 
226
 
227
  - πŸ”₯πŸ”₯πŸ”₯ [7/25/2023] We released **WizardLM V1.2** models. The **WizardLM-13B-V1.2** is here ([Demo_13B-V1.2](https://b7a19878988c8c73.gradio.app), [Demo_13B-V1.2_bak-1](https://d0a37a76e0ac4b52.gradio.app/), [Full Model Weight](https://huggingface.co/WizardLM/WizardLM-13B-V1.2)). Please checkout the [paper](https://arxiv.org/abs/2304.12244).
228
  - πŸ”₯πŸ”₯πŸ”₯ [7/25/2023] The **WizardLM-13B-V1.2** achieves **7.06** on [MT-Bench Leaderboard](https://chat.lmsys.org/?leaderboard), **89.17%** on [AlpacaEval Leaderboard](https://tatsu-lab.github.io/alpaca_eval/), and **101.4%** on [WizardLM Eval](https://github.com/nlpxucan/WizardLM/blob/main/WizardLM/data/WizardLM_testset.jsonl). (Note: MT-Bench and AlpacaEval are all self-test, will push update and request review. All tests are completed under their official settings.)
229
+
230
+ ❗<b>Note for model system prompts usage:</b>
231
+
232
+
233
+ <b>WizardLM</b> adopts the prompt format from <b>Vicuna</b> and supports **multi-turn** conversation. The prompt should be as following:
234
+
235
+ ```
236
+ A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: Hi ASSISTANT: Hello.</s>USER: Who are you? ASSISTANT: I am WizardLM.</s>......
237
+ ```
238
+
239
+ ## Inference WizardLM Demo Script
240
+
241
+ We provide the inference WizardLM demo code [here](https://github.com/nlpxucan/WizardLM/tree/main/demo).
242
+
243
+ ❗<b>To commen concern about dataset:</b>
244
+
245
+ Recently, there have been clear changes in the open-source policy and regulations of our overall organization's code, data, and models.
246
+
247
+
248
+ Despite this, we have still worked hard to obtain opening the weights of the model first, but the data involves stricter auditing and is in review with our legal team .
249
+
250
+ Our researchers have no authority to publicly release them without authorization.
251
+
252
+ Thank you for your understanding.