Transformers
llama
TheBloke commited on
Commit
ecb10c9
1 Parent(s): 12d4401

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +50 -21
README.md CHANGED
@@ -11,17 +11,20 @@ quantized_by: TheBloke
11
  ---
12
 
13
  <!-- header start -->
14
- <div style="width: 100%;">
15
- <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
 
16
  </div>
17
  <div style="display: flex; justify-content: space-between; width: 100%;">
18
  <div style="display: flex; flex-direction: column; align-items: flex-start;">
19
- <p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
20
  </div>
21
  <div style="display: flex; flex-direction: column; align-items: flex-end;">
22
- <p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
23
  </div>
24
  </div>
 
 
25
  <!-- header end -->
26
 
27
  # LosslessMegaCoder Llama2 13B Mini - GGML
@@ -32,6 +35,13 @@ quantized_by: TheBloke
32
 
33
  This repo contains GGML format model files for [Rombo Dawg's LosslessMegaCoder Llama2 13B Mini](https://huggingface.co/rombodawg/LosslessMegaCoder-llama2-13b-mini).
34
 
 
 
 
 
 
 
 
35
  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:
36
  * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most popular web UI. Supports NVidia CUDA GPU acceleration.
37
  * [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.
@@ -43,7 +53,8 @@ GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/gger
43
  ## Repositories available
44
 
45
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GPTQ)
46
- * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML)
 
47
  * [Rombo Dawg's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/rombodawg/LosslessMegaCoder-llama2-13b-mini)
48
 
49
  ## Prompt template: ChatML
@@ -54,14 +65,19 @@ GGML files are for CPU + GPU inference using [llama.cpp](https://github.com/gger
54
  <|im_start|>user
55
  {prompt}<|im_end|>
56
  <|im_start|>assistant
 
57
  ```
58
 
59
  <!-- compatibility_ggml start -->
60
  ## Compatibility
61
 
62
- These quantised GGML files are compatible with llama.cpp as of June 6th, commit `2d43387`.
 
 
 
 
63
 
64
- They should also be compatible with all UIs, libraries and utilities which use GGML.
65
 
66
  ## Explanation of the new k-quant methods
67
  <details>
@@ -84,17 +100,17 @@ Refer to the Provided Files table below to see what files use which methods, and
84
  | Name | Quant method | Bits | Size | Max RAM required | Use case |
85
  | ---- | ---- | ---- | ---- | ---- | ----- |
86
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q2_K.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q2_K.bin) | q2_K | 2 | 5.74 GB| 8.24 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. |
87
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_L.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_L.bin) | q3_K_L | 3 | 7.14 GB| 9.64 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 |
88
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_M.bin) | q3_K_M | 3 | 6.53 GB| 9.03 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 |
89
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_S.bin) | q3_K_S | 3 | 5.87 GB| 8.37 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
 
 
90
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_0.bin) | q4_0 | 4 | 7.32 GB| 9.82 GB | Original quant method, 4-bit. |
91
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_1.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
92
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin) | q4_K_M | 4 | 8.06 GB| 10.56 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 |
93
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_S.bin) | q4_K_S | 4 | 7.56 GB| 10.06 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
 
 
94
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
95
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_1.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
96
- | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_M.bin) | q5_K_M | 5 | 9.40 GB| 11.90 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 |
97
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_S.bin) | q5_K_S | 5 | 9.14 GB| 11.64 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
 
 
98
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q6_K.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q6_K.bin) | q6_K | 6 | 10.83 GB| 13.33 GB | New k-quant method. Uses GGML_TYPE_Q8_K for all tensors - 6-bit quantization |
99
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q8_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
100
 
@@ -102,10 +118,12 @@ Refer to the Provided Files table below to see what files use which methods, and
102
 
103
  ## How to run in `llama.cpp`
104
 
105
- I use the following command line; adjust for your tastes and needs:
 
 
106
 
107
  ```
108
- ./main -t 10 -ngl 32 -m losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: Write a story about llamas\n### Response:"
109
  ```
110
  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`.
111
 
@@ -122,6 +140,7 @@ For other parameters and how to use them, please refer to [the llama.cpp documen
122
  Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
123
 
124
  <!-- footer start -->
 
125
  ## Discord
126
 
127
  For further support, and discussions on these models and AI in general, join us at:
@@ -141,18 +160,24 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
141
  * Patreon: https://patreon.com/TheBlokeAI
142
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
143
 
144
- **Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
145
 
146
- **Patreon special mentions**: Ajan Kanaga, David Ziegler, Raymond Fosdick, SuperWojo, Sam, webtim, Steven Wood, knownsqashed, Tony Hughes, Junyu Yang, J, Olakabola, Dan Guido, Stephen Murray, John Villwock, vamX, William Sang, Sean Connelly, LangChain4j, Olusegun Samson, Fen Risland, Derek Yates, Karl Bernard, transmissions 11, Trenton Dambrowitz, Pieter, Preetika Verma, Swaroop Kallakuri, Andrey, Slarti, Jonathan Leane, Michael Levine, Kalila, Joseph William Delisle, Rishabh Srivastava, Deo Leter, Luke Pendergrass, Spencer Kim, Geoffrey Montalvo, Thomas Belote, Jeffrey Morgan, Mandus, ya boyyy, Matthew Berman, Magnesian, Ai Maven, senxiiz, Alps Aficionado, Luke @flexchar, Raven Klaugh, Imad Khwaja, Gabriel Puliatti, Johann-Peter Hartmann, usrbinkat, Spiking Neurons AB, Artur Olbinski, chris gileta, danny, Willem Michiel, WelcomeToTheClub, Deep Realms, alfie_i, Dave, Leonard Tan, NimbleBox.ai, Randy H, Daniel P. Andersen, Pyrater, Will Dee, Elle, Space Cruiser, Gabriel Tamborski, Asp the Wyvern, Illia Dulskyi, Nikolai Manek, Sid, Brandon Frisco, Nathan LeClaire, Edmond Seymore, Enrico Ros, Pedro Madruga, Eugene Pentland, John Detwiler, Mano Prime, Stanislav Ovsiannikov, Alex, Vitor Caleffi, K, biorpg, Michael Davis, Lone Striker, Pierre Kircher, theTransient, Fred von Graf, Sebastain Graf, Vadim, Iucharbius, Clay Pascal, Chadd, Mesiah Bishop, terasurfer, Rainer Wilmers, Alexandros Triantafyllidis, Stefan Sabev, Talal Aujan, Cory Kujawski, Viktor Bowallius, subjectnull, ReadyPlayerEmma, zynix
147
 
148
 
149
  Thank you to all my generous patrons and donaters!
150
 
 
 
151
  <!-- footer end -->
152
 
153
  # Original model card: Rombo Dawg's LosslessMegaCoder Llama2 13B Mini
154
 
155
- This is one of the first models trained on the LosslessMegaCodeTrainingV2_1m_Evol_Uncensored dataset. The version of the dataset used for this model was poorly filtered on some loose parameters that arent anything to write home about but plans for much more refined filtering are in the works
 
 
 
 
156
 
157
  - This model was made as a colaboration between me and andreaskoepf who is an affiliate of Open Assistant.
158
 
@@ -178,13 +203,13 @@ Gpt4all template:
178
  - System prompt
179
  ```
180
  <|im_start|>system
181
- {system message}
182
  ```
183
  - Prompt template
184
  ```
185
  <|im_end|>
186
  <|im_start|>user
187
- %1<|im_end|>
188
  <|im_start|>assistant
189
  ```
190
 
@@ -209,6 +234,10 @@ Oobagooba Text-Generation-Webui Template
209
  Below is an instruction that describes a task. Write a response that appropriately completes the request.<|im_end|>
210
  ```
211
 
 
 
 
 
212
  Training data:
213
 
214
  - https://wandb.ai/open-assistant/epfl-mt-sft/runs/run34_megacode2_min100_13b
@@ -223,4 +252,4 @@ Link for the filtered dataset used to make this model are bellow:
223
 
224
  The original posting for this model was uploaded at the link bellow.
225
 
226
- - https://huggingface.co/andreaskoepf/llama2-13b-megacode2_min100
 
11
  ---
12
 
13
  <!-- header start -->
14
+ <!-- 200823 -->
15
+ <div style="width: auto; margin-left: auto; margin-right: auto">
16
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
17
  </div>
18
  <div style="display: flex; justify-content: space-between; width: 100%;">
19
  <div style="display: flex; flex-direction: column; align-items: flex-start;">
20
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
21
  </div>
22
  <div style="display: flex; flex-direction: column; align-items: flex-end;">
23
+ <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>
24
  </div>
25
  </div>
26
+ <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>
27
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
28
  <!-- header end -->
29
 
30
  # LosslessMegaCoder Llama2 13B Mini - GGML
 
35
 
36
  This repo contains GGML format model files for [Rombo Dawg's LosslessMegaCoder Llama2 13B Mini](https://huggingface.co/rombodawg/LosslessMegaCoder-llama2-13b-mini).
37
 
38
+ ### Important note regarding GGML files.
39
+
40
+ 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.
41
+
42
+ Please use the GGUF models instead.
43
+ ### About GGML
44
+
45
  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:
46
  * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most popular web UI. Supports NVidia CUDA GPU acceleration.
47
  * [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.
 
53
  ## Repositories available
54
 
55
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GPTQ)
56
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGUF)
57
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML)
58
  * [Rombo Dawg's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/rombodawg/LosslessMegaCoder-llama2-13b-mini)
59
 
60
  ## Prompt template: ChatML
 
65
  <|im_start|>user
66
  {prompt}<|im_end|>
67
  <|im_start|>assistant
68
+
69
  ```
70
 
71
  <!-- compatibility_ggml start -->
72
  ## Compatibility
73
 
74
+ These quantised GGML files are compatible with llama.cpp between June 6th (commit `2d43387`) and August 21st 2023.
75
+
76
+ For support with latest llama.cpp, please use GGUF files instead.
77
+
78
+ The final llama.cpp commit with support for GGML was: [dadbed99e65252d79f81101a392d0d6497b86caa](https://github.com/ggerganov/llama.cpp/commit/dadbed99e65252d79f81101a392d0d6497b86caa)
79
 
80
+ As of August 23rd 2023 they are still compatible with all UIs, libraries and utilities which use GGML. This may change in the future.
81
 
82
  ## Explanation of the new k-quant methods
83
  <details>
 
100
  | Name | Quant method | Bits | Size | Max RAM required | Use case |
101
  | ---- | ---- | ---- | ---- | ---- | ----- |
102
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q2_K.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q2_K.bin) | q2_K | 2 | 5.74 GB| 8.24 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. |
 
 
103
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_S.bin) | q3_K_S | 3 | 5.87 GB| 8.37 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
104
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_M.bin) | q3_K_M | 3 | 6.53 GB| 9.03 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 |
105
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_L.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q3_K_L.bin) | q3_K_L | 3 | 7.14 GB| 9.64 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 |
106
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_0.bin) | q4_0 | 4 | 7.32 GB| 9.82 GB | Original quant method, 4-bit. |
 
 
107
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_S.bin) | q4_K_S | 4 | 7.56 GB| 10.06 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
108
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin) | q4_K_M | 4 | 8.06 GB| 10.56 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 |
109
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q4_1.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
110
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
 
 
111
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_S.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_S.bin) | q5_K_S | 5 | 9.14 GB| 11.64 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
112
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_M.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q5_K_M.bin) | q5_K_M | 5 | 9.40 GB| 11.90 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 |
113
+ | [losslessmegacoder-llama2-13b-min.ggmlv3.q5_1.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
114
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q6_K.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.ggmlv3.q6_K.bin) | q6_K | 6 | 10.83 GB| 13.33 GB | New k-quant method. Uses GGML_TYPE_Q8_K for all tensors - 6-bit quantization |
115
  | [losslessmegacoder-llama2-13b-min.ggmlv3.q8_0.bin](https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GGML/blob/main/losslessmegacoder-llama2-13b-min.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. |
116
 
 
118
 
119
  ## How to run in `llama.cpp`
120
 
121
+ Make sure you are using `llama.cpp` from commit [dadbed99e65252d79f81101a392d0d6497b86caa](https://github.com/ggerganov/llama.cpp/commit/dadbed99e65252d79f81101a392d0d6497b86caa) or earlier.
122
+
123
+ For compatibility with latest llama.cpp, please use GGUF files instead.
124
 
125
  ```
126
+ ./main -t 10 -ngl 32 -m losslessmegacoder-llama2-13b-min.ggmlv3.q4_K_M.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"
127
  ```
128
  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`.
129
 
 
140
  Further instructions here: [text-generation-webui/docs/llama.cpp.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp.md).
141
 
142
  <!-- footer start -->
143
+ <!-- 200823 -->
144
  ## Discord
145
 
146
  For further support, and discussions on these models and AI in general, join us at:
 
160
  * Patreon: https://patreon.com/TheBlokeAI
161
  * Ko-Fi: https://ko-fi.com/TheBlokeAI
162
 
163
+ **Special thanks to**: Aemon Algiz.
164
 
165
+ **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 Wikieł, 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 Bjäreholt, 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
166
 
167
 
168
  Thank you to all my generous patrons and donaters!
169
 
170
+ And thank you again to a16z for their generous grant.
171
+
172
  <!-- footer end -->
173
 
174
  # Original model card: Rombo Dawg's LosslessMegaCoder Llama2 13B Mini
175
 
176
+ ___________________________
177
+ - Please note this model was not trained on the rombodawg/LosslessMegaCodeTrainingV3_MINI dataset, despite the name similarity. You can find the training data at the bottom of the model card labeled (megacode2-min100)
178
+ ___________________________
179
+
180
+ This is one of the first models trained on the LosslessMegaCodeTrainingV2_1m_Evol_Uncensored dataset. The version of the dataset used for this model was filtered by removed any data with less than 100 tokens but plans for much more refined filtering are in the works
181
 
182
  - This model was made as a colaboration between me and andreaskoepf who is an affiliate of Open Assistant.
183
 
 
203
  - System prompt
204
  ```
205
  <|im_start|>system
206
+ "Below is an instruction that describes a task. Write a response that appropriately completes the request."
207
  ```
208
  - Prompt template
209
  ```
210
  <|im_end|>
211
  <|im_start|>user
212
+ "%1"<|im_end|>
213
  <|im_start|>assistant
214
  ```
215
 
 
234
  Below is an instruction that describes a task. Write a response that appropriately completes the request.<|im_end|>
235
  ```
236
 
237
+ Current quantizations available:
238
+
239
+ - https://huggingface.co/TheBloke/LosslessMegaCoder-Llama2-13B-Mini-GPTQ
240
+
241
  Training data:
242
 
243
  - https://wandb.ai/open-assistant/epfl-mt-sft/runs/run34_megacode2_min100_13b
 
252
 
253
  The original posting for this model was uploaded at the link bellow.
254
 
255
+ - https://huggingface.co/andreaskoepf/llama2-13b-megacode2_min100