TheBloke commited on
Commit
e6e07d5
1 Parent(s): f14d3df

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +71 -48
README.md CHANGED
@@ -2,7 +2,7 @@
2
  datasets:
3
  - totally-not-an-llm/EverythingLM-data
4
  inference: false
5
- license: other
6
  model_creator: Kai Howard
7
  model_link: https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k
8
  model_name: EverythingLM 13B 16K
@@ -31,18 +31,24 @@ quantized_by: TheBloke
31
  - Model creator: [Kai Howard](https://huggingface.co/totally-not-an-llm)
32
  - Original model: [EverythingLM 13B 16K](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k)
33
 
 
34
  ## Description
35
 
36
  This repo contains GPTQ model files for [Kai Howard's EverythingLM 13B 16K](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k).
37
 
38
  Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
39
 
 
 
40
  ## Repositories available
41
 
42
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ)
43
- * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GGML)
 
44
  * [Kai Howard's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k)
 
45
 
 
46
  ## Prompt template: Vicuna-Short
47
 
48
  ```
@@ -50,22 +56,26 @@ You are a helpful AI assistant.
50
 
51
  USER: {prompt}
52
  ASSISTANT:
 
53
  ```
54
 
 
 
 
55
  ## Provided files and GPTQ parameters
56
 
57
  Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
58
 
59
  Each separate quant is in a different branch. See below for instructions on fetching from different branches.
60
 
61
- All GPTQ files are made with AutoGPTQ.
62
 
63
  <details>
64
  <summary>Explanation of GPTQ parameters</summary>
65
 
66
  - Bits: The bit size of the quantised model.
67
  - GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
68
- - Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have issues with models that use Act Order plus Group Size.
69
  - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
70
  - GPTQ dataset: The dataset used for quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
71
  - Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
@@ -75,13 +85,16 @@ All GPTQ files are made with AutoGPTQ.
75
 
76
  | Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
77
  | ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
78
- | [main](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/main) | 4 | 128 | No | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.26 GB | Yes | Most compatible option. Good inference speed in AutoGPTQ and GPTQ-for-LLaMa. Lower inference quality than other options. |
79
- | [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 8.00 GB | Yes | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. Poor AutoGPTQ CUDA speed. |
80
- | [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-64g-actorder_True) | 4 | 64 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.51 GB | Yes | 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed. |
81
- | [gptq-4bit-128g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-128g-actorder_True) | 4 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.26 GB | Yes | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed. |
82
- | [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 13.36 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements and to improve AutoGPTQ speed. |
83
  | [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 13.65 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. Poor AutoGPTQ CUDA speed. |
84
 
 
 
 
85
  ## How to download from branches
86
 
87
  - In text-generation-webui, you can add `:branch` to the end of the download name, eg `TheBloke/EverythingLM-13B-16K-GPTQ:gptq-4bit-32g-actorder_True`
@@ -90,78 +103,78 @@ All GPTQ files are made with AutoGPTQ.
90
  git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ
91
  ```
92
  - In Python Transformers code, the branch is the `revision` parameter; see below.
93
-
 
94
  ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
95
 
96
  Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
97
 
98
- It is strongly recommended to use the text-generation-webui one-click-installers unless you know how to make a manual install.
99
 
100
  1. Click the **Model tab**.
101
  2. Under **Download custom model or LoRA**, enter `TheBloke/EverythingLM-13B-16K-GPTQ`.
102
  - To download from a specific branch, enter for example `TheBloke/EverythingLM-13B-16K-GPTQ:gptq-4bit-32g-actorder_True`
103
  - see Provided Files above for the list of branches for each option.
104
  3. Click **Download**.
105
- 4. The model will start downloading. Once it's finished it will say "Done"
106
  5. In the top left, click the refresh icon next to **Model**.
107
  6. In the **Model** dropdown, choose the model you just downloaded: `EverythingLM-13B-16K-GPTQ`
108
  7. The model will automatically load, and is now ready for use!
109
  8. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
110
- * Note that you do not need to set GPTQ parameters any more. These are set automatically from the file `quantize_config.json`.
111
  9. Once you're ready, click the **Text Generation tab** and enter a prompt to get started!
 
112
 
 
113
  ## How to use this GPTQ model from Python code
114
 
115
- First make sure you have [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ) 0.3.1 or later installed:
116
 
117
- ```
118
- pip3 install auto-gptq
119
- ```
120
 
121
- If you have problems installing AutoGPTQ, please build from source instead:
 
 
122
  ```
 
 
 
 
123
  pip3 uninstall -y auto-gptq
124
  git clone https://github.com/PanQiWei/AutoGPTQ
125
  cd AutoGPTQ
126
  pip3 install .
127
  ```
128
 
129
- Then try the following example code:
 
 
 
 
 
 
 
 
130
 
131
  ```python
132
- from transformers import AutoTokenizer, pipeline, logging
133
- from auto_gptq import AutoGPTQForCausalLM, BaseQuantizeConfig
134
 
135
  model_name_or_path = "TheBloke/EverythingLM-13B-16K-GPTQ"
136
-
137
- use_triton = False
 
 
 
 
138
 
139
  tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
140
 
141
- model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
142
- use_safetensors=True,
143
- trust_remote_code=False,
144
- device="cuda:0",
145
- use_triton=use_triton,
146
- quantize_config=None)
147
-
148
- """
149
- # To download from a specific branch, use the revision parameter, as in this example:
150
- # Note that `revision` requires AutoGPTQ 0.3.1 or later!
151
-
152
- model = AutoGPTQForCausalLM.from_quantized(model_name_or_path,
153
- revision="gptq-4bit-32g-actorder_True",
154
- use_safetensors=True,
155
- trust_remote_code=False,
156
- device="cuda:0",
157
- quantize_config=None)
158
- """
159
-
160
  prompt = "Tell me about AI"
161
  prompt_template=f'''You are a helpful AI assistant.
162
 
163
  USER: {prompt}
164
  ASSISTANT:
 
165
  '''
166
 
167
  print("\n\n*** Generate:")
@@ -172,9 +185,6 @@ print(tokenizer.decode(output[0]))
172
 
173
  # Inference can also be done using transformers' pipeline
174
 
175
- # Prevent printing spurious transformers error when using pipeline with AutoGPTQ
176
- logging.set_verbosity(logging.CRITICAL)
177
-
178
  print("*** Pipeline:")
179
  pipe = pipeline(
180
  "text-generation",
@@ -188,12 +198,17 @@ pipe = pipeline(
188
 
189
  print(pipe(prompt_template)[0]['generated_text'])
190
  ```
 
191
 
 
192
  ## Compatibility
193
 
194
- The files provided will work with AutoGPTQ (CUDA and Triton modes), GPTQ-for-LLaMa (only CUDA has been tested), and Occ4m's GPTQ-for-LLaMa fork.
 
 
195
 
196
- ExLlama works with Llama models in 4-bit. Please see the Provided Files table above for per-file compatibility.
 
197
 
198
  <!-- footer start -->
199
  <!-- 200823 -->
@@ -218,7 +233,7 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
218
 
219
  **Special thanks to**: Aemon Algiz.
220
 
221
- **Patreon special mentions**: Sam, theTransient, Jonathan Leane, Steven Wood, webtim, Johann-Peter Hartmann, Geoffrey Montalvo, Gabriel Tamborski, Willem Michiel, John Villwock, Derek Yates, Mesiah Bishop, Eugene Pentland, Pieter, Chadd, Stephen Murray, Daniel P. Andersen, terasurfer, Brandon Frisco, Thomas Belote, Sid, Nathan LeClaire, Magnesian, Alps Aficionado, Stanislav Ovsiannikov, Alex, Joseph William Delisle, Nikolai Manek, Michael Davis, Junyu Yang, K, J, Spencer Kim, Stefan Sabev, Olusegun Samson, transmissions 11, Michael Levine, Cory Kujawski, Rainer Wilmers, zynix, Kalila, Luke @flexchar, Ajan Kanaga, Mandus, vamX, Ai Maven, Mano Prime, Matthew Berman, subjectnull, Vitor Caleffi, Clay Pascal, biorpg, alfie_i, 阿明, Jeffrey Morgan, ya boyyy, Raymond Fosdick, knownsqashed, Olakabola, Leonard Tan, ReadyPlayerEmma, Enrico Ros, Dave, Talal Aujan, Illia Dulskyi, Sean Connelly, senxiiz, Artur Olbinski, Elle, Raven Klaugh, Fen Risland, Deep Realms, Imad Khwaja, Fred von Graf, Will Dee, usrbinkat, SuperWojo, Alexandros Triantafyllidis, Swaroop Kallakuri, Dan Guido, John Detwiler, Pedro Madruga, Iucharbius, Viktor Bowallius, Asp the Wyvern, Edmond Seymore, Trenton Dambrowitz, Space Cruiser, Spiking Neurons AB, Pyrater, LangChain4j, Tony Hughes, Kacper Wikieł, Rishabh Srivastava, David Ziegler, Luke Pendergrass, Andrey, Gabriel Puliatti, Lone Striker, Sebastain Graf, Pierre Kircher, Randy H, NimbleBox.ai, Vadim, danny, Deo Leter
222
 
223
 
224
  Thank you to all my generous patrons and donaters!
@@ -238,6 +253,14 @@ The model is completely uncensored.
238
 
239
  This model is an early test of the EverythingLM dataset and some new experimental principles, so don't consider it SOTA.
240
 
 
 
 
 
 
 
 
 
241
  ### Notable features:
242
  - Automatically triggered CoT reasoning.
243
  - Verbose and detailed replies.
 
2
  datasets:
3
  - totally-not-an-llm/EverythingLM-data
4
  inference: false
5
+ license: llama2
6
  model_creator: Kai Howard
7
  model_link: https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k
8
  model_name: EverythingLM 13B 16K
 
31
  - Model creator: [Kai Howard](https://huggingface.co/totally-not-an-llm)
32
  - Original model: [EverythingLM 13B 16K](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k)
33
 
34
+ <!-- description start -->
35
  ## Description
36
 
37
  This repo contains GPTQ model files for [Kai Howard's EverythingLM 13B 16K](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k).
38
 
39
  Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
40
 
41
+ <!-- description end -->
42
+ <!-- repositories-available start -->
43
  ## Repositories available
44
 
45
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ)
46
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GGUF)
47
+ * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GGML)
48
  * [Kai Howard's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/totally-not-an-llm/EverythingLM-13b-16k)
49
+ <!-- repositories-available end -->
50
 
51
+ <!-- prompt-template start -->
52
  ## Prompt template: Vicuna-Short
53
 
54
  ```
 
56
 
57
  USER: {prompt}
58
  ASSISTANT:
59
+
60
  ```
61
 
62
+ <!-- prompt-template end -->
63
+
64
+ <!-- README_GPTQ.md-provided-files start -->
65
  ## Provided files and GPTQ parameters
66
 
67
  Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
68
 
69
  Each separate quant is in a different branch. See below for instructions on fetching from different branches.
70
 
71
+ All recent GPTQ files are made with AutoGPTQ, and all files in non-main branches are made with AutoGPTQ. Files in the `main` branch which were uploaded before August 2023 were made with GPTQ-for-LLaMa.
72
 
73
  <details>
74
  <summary>Explanation of GPTQ parameters</summary>
75
 
76
  - Bits: The bit size of the quantised model.
77
  - GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
78
+ - Act Order: True or False. Also known as `desc_act`. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now.
79
  - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
80
  - GPTQ dataset: The dataset used for quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
81
  - Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
 
85
 
86
  | Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
87
  | ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
88
+ | [main](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/main) | 4 | 128 | No | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.26 GB | Yes | Most compatible option. Good inference speed in AutoGPTQ and GPTQ-for-LLaMa. Lower inference quality than other options. |
89
+ | [gptq-4bit-32g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-32g-actorder_True) | 4 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 8.00 GB | Yes | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. Poor AutoGPTQ CUDA speed. |
90
+ | [gptq-4bit-64g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-64g-actorder_True) | 4 | 64 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.51 GB | Yes | 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed. |
91
+ | [gptq-4bit-128g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-4bit-128g-actorder_True) | 4 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 7.26 GB | Yes | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. Poor AutoGPTQ CUDA speed. |
92
+ | [gptq-8bit--1g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-8bit--1g-actorder_True) | 8 | None | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 13.36 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements and to improve AutoGPTQ speed. |
93
  | [gptq-8bit-128g-actorder_True](https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ/tree/gptq-8bit-128g-actorder_True) | 8 | 128 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 13.65 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. Poor AutoGPTQ CUDA speed. |
94
 
95
+ <!-- README_GPTQ.md-provided-files end -->
96
+
97
+ <!-- README_GPTQ.md-download-from-branches start -->
98
  ## How to download from branches
99
 
100
  - In text-generation-webui, you can add `:branch` to the end of the download name, eg `TheBloke/EverythingLM-13B-16K-GPTQ:gptq-4bit-32g-actorder_True`
 
103
  git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ
104
  ```
105
  - In Python Transformers code, the branch is the `revision` parameter; see below.
106
+ <!-- README_GPTQ.md-download-from-branches end -->
107
+ <!-- README_GPTQ.md-text-generation-webui start -->
108
  ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
109
 
110
  Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
111
 
112
+ It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.
113
 
114
  1. Click the **Model tab**.
115
  2. Under **Download custom model or LoRA**, enter `TheBloke/EverythingLM-13B-16K-GPTQ`.
116
  - To download from a specific branch, enter for example `TheBloke/EverythingLM-13B-16K-GPTQ:gptq-4bit-32g-actorder_True`
117
  - see Provided Files above for the list of branches for each option.
118
  3. Click **Download**.
119
+ 4. The model will start downloading. Once it's finished it will say "Done".
120
  5. In the top left, click the refresh icon next to **Model**.
121
  6. In the **Model** dropdown, choose the model you just downloaded: `EverythingLM-13B-16K-GPTQ`
122
  7. The model will automatically load, and is now ready for use!
123
  8. If you want any custom settings, set them and then click **Save settings for this model** followed by **Reload the Model** in the top right.
124
+ * Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file `quantize_config.json`.
125
  9. Once you're ready, click the **Text Generation tab** and enter a prompt to get started!
126
+ <!-- README_GPTQ.md-text-generation-webui end -->
127
 
128
+ <!-- README_GPTQ.md-use-from-python start -->
129
  ## How to use this GPTQ model from Python code
130
 
131
+ ### Install the necessary packages
132
 
133
+ Requires: Transformers 4.32.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
 
 
134
 
135
+ ```shell
136
+ pip3 install transformers>=4.32.0 optimum>=1.12.0
137
+ pip3 install auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/ # Use cu117 if on CUDA 11.7
138
  ```
139
+
140
+ If you have problems installing AutoGPTQ using the pre-built wheels, install it from source instead:
141
+
142
+ ```shell
143
  pip3 uninstall -y auto-gptq
144
  git clone https://github.com/PanQiWei/AutoGPTQ
145
  cd AutoGPTQ
146
  pip3 install .
147
  ```
148
 
149
+ ### For CodeLlama models only: you must use Transformers 4.33.0 or later.
150
+
151
+ If 4.33.0 is not yet released when you read this, you will need to install Transformers from source:
152
+ ```shell
153
+ pip3 uninstall -y transformers
154
+ pip3 install git+https://github.com/huggingface/transformers.git
155
+ ```
156
+
157
+ ### You can then use the following code
158
 
159
  ```python
160
+ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 
161
 
162
  model_name_or_path = "TheBloke/EverythingLM-13B-16K-GPTQ"
163
+ # To use a different branch, change revision
164
+ # For example: revision="gptq-4bit-32g-actorder_True"
165
+ model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
166
+ torch_dtype=torch.float16,
167
+ device_map="auto",
168
+ revision="main")
169
 
170
  tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
171
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
172
  prompt = "Tell me about AI"
173
  prompt_template=f'''You are a helpful AI assistant.
174
 
175
  USER: {prompt}
176
  ASSISTANT:
177
+
178
  '''
179
 
180
  print("\n\n*** Generate:")
 
185
 
186
  # Inference can also be done using transformers' pipeline
187
 
 
 
 
188
  print("*** Pipeline:")
189
  pipe = pipeline(
190
  "text-generation",
 
198
 
199
  print(pipe(prompt_template)[0]['generated_text'])
200
  ```
201
+ <!-- README_GPTQ.md-use-from-python end -->
202
 
203
+ <!-- README_GPTQ.md-compatibility start -->
204
  ## Compatibility
205
 
206
+ The files provided are tested to work with AutoGPTQ, both via Transformers and using AutoGPTQ directly. They should also work with [Occ4m's GPTQ-for-LLaMa fork](https://github.com/0cc4m/KoboldAI).
207
+
208
+ [ExLlama](https://github.com/turboderp/exllama) is compatible with Llama models in 4-bit. Please see the Provided Files table above for per-file compatibility.
209
 
210
+ [Huggingface Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) is compatible with all GPTQ models.
211
+ <!-- README_GPTQ.md-compatibility end -->
212
 
213
  <!-- footer start -->
214
  <!-- 200823 -->
 
233
 
234
  **Special thanks to**: Aemon Algiz.
235
 
236
+ **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
237
 
238
 
239
  Thank you to all my generous patrons and donaters!
 
253
 
254
  This model is an early test of the EverythingLM dataset and some new experimental principles, so don't consider it SOTA.
255
 
256
+ ### GGML quants:
257
+ https://huggingface.co/TheBloke/EverythingLM-13B-16K-GGML
258
+
259
+ Make sure to use correct rope scaling settings:
260
+ `-c 16384 --rope-freq-base 10000 --rope-freq-scale 0.25`
261
+ ### GPTQ quants:
262
+ https://huggingface.co/TheBloke/EverythingLM-13B-16K-GPTQ
263
+
264
  ### Notable features:
265
  - Automatically triggered CoT reasoning.
266
  - Verbose and detailed replies.