TheBloke commited on
Commit
fec3254
1 Parent(s): 7049375

Upload README.md

Browse files
Files changed (1) hide show
  1. README.md +420 -0
README.md ADDED
@@ -0,0 +1,420 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: ajibawa-2023/Python-Code-13B
3
+ datasets:
4
+ - ajibawa-2023/Python-Code-23k-ShareGPT
5
+ inference: false
6
+ language:
7
+ - en
8
+ license: cc-by-nc-nd-4.0
9
+ model_creator: Feynman Innovations
10
+ model_name: Python Code 13B
11
+ model_type: llama
12
+ prompt_template: 'This is a conversation with your helpful AI assistant. AI assistant
13
+ can generate Python Code along with necessary explanation.
14
+
15
+
16
+ Context
17
+
18
+ You are a helpful AI assistant.
19
+
20
+
21
+ USER: {prompt}
22
+
23
+ ASSISTANT:
24
+
25
+ '
26
+ quantized_by: TheBloke
27
+ tags:
28
+ - code
29
+ ---
30
+ <!-- markdownlint-disable MD041 -->
31
+
32
+ <!-- header start -->
33
+ <!-- 200823 -->
34
+ <div style="width: auto; margin-left: auto; margin-right: auto">
35
+ <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
36
+ </div>
37
+ <div style="display: flex; justify-content: space-between; width: 100%;">
38
+ <div style="display: flex; flex-direction: column; align-items: flex-start;">
39
+ <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
40
+ </div>
41
+ <div style="display: flex; flex-direction: column; align-items: flex-end;">
42
+ <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>
43
+ </div>
44
+ </div>
45
+ <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>
46
+ <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
47
+ <!-- header end -->
48
+
49
+ # Python Code 13B - AWQ
50
+ - Model creator: [Feynman Innovations](https://huggingface.co/ajibawa-2023)
51
+ - Original model: [Python Code 13B](https://huggingface.co/ajibawa-2023/Python-Code-13B)
52
+
53
+ <!-- description start -->
54
+ ## Description
55
+
56
+ This repo contains AWQ model files for [Feynman Innovations's Python Code 13B](https://huggingface.co/ajibawa-2023/Python-Code-13B).
57
+
58
+ These files were quantised using hardware kindly provided by [Massed Compute](https://massedcompute.com/).
59
+
60
+
61
+ ### About AWQ
62
+
63
+ AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.
64
+
65
+ It is supported by:
66
+
67
+ - [Text Generation Webui](https://github.com/oobabooga/text-generation-webui) - using Loader: AutoAWQ
68
+ - [vLLM](https://github.com/vllm-project/vllm) - Llama and Mistral models only
69
+ - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference)
70
+ - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later, from any code or client that supports Transformers
71
+ - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) - for use from Python code
72
+
73
+ <!-- description end -->
74
+ <!-- repositories-available start -->
75
+ ## Repositories available
76
+
77
+ * [AWQ model(s) for GPU inference.](https://huggingface.co/TheBloke/Python-Code-13B-AWQ)
78
+ * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/Python-Code-13B-GPTQ)
79
+ * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/Python-Code-13B-GGUF)
80
+ * [Feynman Innovations's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ajibawa-2023/Python-Code-13B)
81
+ <!-- repositories-available end -->
82
+
83
+ <!-- prompt-template start -->
84
+ ## Prompt template: Ajibawa-Python-Code
85
+
86
+ ```
87
+ This is a conversation with your helpful AI assistant. AI assistant can generate Python Code along with necessary explanation.
88
+
89
+ Context
90
+ You are a helpful AI assistant.
91
+
92
+ USER: {prompt}
93
+ ASSISTANT:
94
+
95
+ ```
96
+
97
+ <!-- prompt-template end -->
98
+ <!-- licensing start -->
99
+ ## Licensing
100
+
101
+ The creator of the source model has listed its license as `cc-by-nc-nd-4.0`, and this quantization has therefore used that same license.
102
+
103
+ As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
104
+
105
+ In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: [Feynman Innovations's Python Code 13B](https://huggingface.co/ajibawa-2023/Python-Code-13B).
106
+ <!-- licensing end -->
107
+ <!-- README_AWQ.md-provided-files start -->
108
+ ## Provided files, and AWQ parameters
109
+
110
+ I currently release 128g GEMM models only. The addition of group_size 32 models, and GEMV kernel models, is being actively considered.
111
+
112
+ Models are released as sharded safetensors files.
113
+
114
+ | Branch | Bits | GS | AWQ Dataset | Seq Len | Size |
115
+ | ------ | ---- | -- | ----------- | ------- | ---- |
116
+ | [main](https://huggingface.co/TheBloke/Python-Code-13B-AWQ/tree/main) | 4 | 128 | [Evol Instruct Code](https://huggingface.co/datasets/nickrosh/Evol-Instruct-Code-80k-v1) | 4096 | 7.25 GB
117
+
118
+ <!-- README_AWQ.md-provided-files end -->
119
+
120
+ <!-- README_AWQ.md-text-generation-webui start -->
121
+ ## How to easily download and use this model in [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
122
+
123
+ Please make sure you're using the latest version of [text-generation-webui](https://github.com/oobabooga/text-generation-webui).
124
+
125
+ 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.
126
+
127
+ 1. Click the **Model tab**.
128
+ 2. Under **Download custom model or LoRA**, enter `TheBloke/Python-Code-13B-AWQ`.
129
+ 3. Click **Download**.
130
+ 4. The model will start downloading. Once it's finished it will say "Done".
131
+ 5. In the top left, click the refresh icon next to **Model**.
132
+ 6. In the **Model** dropdown, choose the model you just downloaded: `Python-Code-13B-AWQ`
133
+ 7. Select **Loader: AutoAWQ**.
134
+ 8. Click Load, and the model will load and is now ready for use.
135
+ 9. 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.
136
+ 10. Once you're ready, click the **Text Generation** tab and enter a prompt to get started!
137
+ <!-- README_AWQ.md-text-generation-webui end -->
138
+
139
+ <!-- README_AWQ.md-use-from-vllm start -->
140
+ ## Multi-user inference server: vLLM
141
+
142
+ Documentation on installing and using vLLM [can be found here](https://vllm.readthedocs.io/en/latest/).
143
+
144
+ - Please ensure you are using vLLM version 0.2 or later.
145
+ - When using vLLM as a server, pass the `--quantization awq` parameter.
146
+
147
+ For example:
148
+
149
+ ```shell
150
+ python3 -m vllm.entrypoints.api_server --model TheBloke/Python-Code-13B-AWQ --quantization awq --dtype auto
151
+ ```
152
+
153
+ - When using vLLM from Python code, again set `quantization=awq`.
154
+
155
+ For example:
156
+
157
+ ```python
158
+ from vllm import LLM, SamplingParams
159
+
160
+ prompts = [
161
+ "Tell me about AI",
162
+ "Write a story about llamas",
163
+ "What is 291 - 150?",
164
+ "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
165
+ ]
166
+ prompt_template=f'''This is a conversation with your helpful AI assistant. AI assistant can generate Python Code along with necessary explanation.
167
+
168
+ Context
169
+ You are a helpful AI assistant.
170
+
171
+ USER: {prompt}
172
+ ASSISTANT:
173
+ '''
174
+
175
+ prompts = [prompt_template.format(prompt=prompt) for prompt in prompts]
176
+
177
+ sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
178
+
179
+ llm = LLM(model="TheBloke/Python-Code-13B-AWQ", quantization="awq", dtype="auto")
180
+
181
+ outputs = llm.generate(prompts, sampling_params)
182
+
183
+ # Print the outputs.
184
+ for output in outputs:
185
+ prompt = output.prompt
186
+ generated_text = output.outputs[0].text
187
+ print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")
188
+ ```
189
+ <!-- README_AWQ.md-use-from-vllm start -->
190
+
191
+ <!-- README_AWQ.md-use-from-tgi start -->
192
+ ## Multi-user inference server: Hugging Face Text Generation Inference (TGI)
193
+
194
+ Use TGI version 1.1.0 or later. The official Docker container is: `ghcr.io/huggingface/text-generation-inference:1.1.0`
195
+
196
+ Example Docker parameters:
197
+
198
+ ```shell
199
+ --model-id TheBloke/Python-Code-13B-AWQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096
200
+ ```
201
+
202
+ Example Python code for interfacing with TGI (requires [huggingface-hub](https://github.com/huggingface/huggingface_hub) 0.17.0 or later):
203
+
204
+ ```shell
205
+ pip3 install huggingface-hub
206
+ ```
207
+
208
+ ```python
209
+ from huggingface_hub import InferenceClient
210
+
211
+ endpoint_url = "https://your-endpoint-url-here"
212
+
213
+ prompt = "Tell me about AI"
214
+ prompt_template=f'''This is a conversation with your helpful AI assistant. AI assistant can generate Python Code along with necessary explanation.
215
+
216
+ Context
217
+ You are a helpful AI assistant.
218
+
219
+ USER: {prompt}
220
+ ASSISTANT:
221
+ '''
222
+
223
+ client = InferenceClient(endpoint_url)
224
+ response = client.text_generation(prompt,
225
+ max_new_tokens=128,
226
+ do_sample=True,
227
+ temperature=0.7,
228
+ top_p=0.95,
229
+ top_k=40,
230
+ repetition_penalty=1.1)
231
+
232
+ print(f"Model output: ", response)
233
+ ```
234
+ <!-- README_AWQ.md-use-from-tgi end -->
235
+
236
+ <!-- README_AWQ.md-use-from-python start -->
237
+ ## Inference from Python code using Transformers
238
+
239
+ ### Install the necessary packages
240
+
241
+ - Requires: [Transformers](https://huggingface.co/docs/transformers) 4.35.0 or later.
242
+ - Requires: [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) 0.1.6 or later.
243
+
244
+ ```shell
245
+ pip3 install --upgrade "autoawq>=0.1.6" "transformers>=4.35.0"
246
+ ```
247
+
248
+ Note that if you are using PyTorch 2.0.1, the above AutoAWQ command will automatically upgrade you to PyTorch 2.1.0.
249
+
250
+ If you are using CUDA 11.8 and wish to continue using PyTorch 2.0.1, instead run this command:
251
+
252
+ ```shell
253
+ pip3 install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp310-cp310-linux_x86_64.whl
254
+ ```
255
+
256
+ If you have problems installing [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) using the pre-built wheels, install it from source instead:
257
+
258
+ ```shell
259
+ pip3 uninstall -y autoawq
260
+ git clone https://github.com/casper-hansen/AutoAWQ
261
+ cd AutoAWQ
262
+ pip3 install .
263
+ ```
264
+
265
+ ### Transformers example code (requires Transformers 4.35.0 and later)
266
+
267
+ ```python
268
+ from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
269
+
270
+ model_name_or_path = "TheBloke/Python-Code-13B-AWQ"
271
+
272
+ tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
273
+ model = AutoModelForCausalLM.from_pretrained(
274
+ model_name_or_path,
275
+ low_cpu_mem_usage=True,
276
+ device_map="cuda:0"
277
+ )
278
+
279
+ # Using the text streamer to stream output one token at a time
280
+ streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
281
+
282
+ prompt = "Tell me about AI"
283
+ prompt_template=f'''This is a conversation with your helpful AI assistant. AI assistant can generate Python Code along with necessary explanation.
284
+
285
+ Context
286
+ You are a helpful AI assistant.
287
+
288
+ USER: {prompt}
289
+ ASSISTANT:
290
+ '''
291
+
292
+ # Convert prompt to tokens
293
+ tokens = tokenizer(
294
+ prompt_template,
295
+ return_tensors='pt'
296
+ ).input_ids.cuda()
297
+
298
+ generation_params = {
299
+ "do_sample": True,
300
+ "temperature": 0.7,
301
+ "top_p": 0.95,
302
+ "top_k": 40,
303
+ "max_new_tokens": 512,
304
+ "repetition_penalty": 1.1
305
+ }
306
+
307
+ # Generate streamed output, visible one token at a time
308
+ generation_output = model.generate(
309
+ tokens,
310
+ streamer=streamer,
311
+ **generation_params
312
+ )
313
+
314
+ # Generation without a streamer, which will include the prompt in the output
315
+ generation_output = model.generate(
316
+ tokens,
317
+ **generation_params
318
+ )
319
+
320
+ # Get the tokens from the output, decode them, print them
321
+ token_output = generation_output[0]
322
+ text_output = tokenizer.decode(token_output)
323
+ print("model.generate output: ", text_output)
324
+
325
+ # Inference is also possible via Transformers' pipeline
326
+ from transformers import pipeline
327
+
328
+ pipe = pipeline(
329
+ "text-generation",
330
+ model=model,
331
+ tokenizer=tokenizer,
332
+ **generation_params
333
+ )
334
+
335
+ pipe_output = pipe(prompt_template)[0]['generated_text']
336
+ print("pipeline output: ", pipe_output)
337
+
338
+ ```
339
+ <!-- README_AWQ.md-use-from-python end -->
340
+
341
+ <!-- README_AWQ.md-compatibility start -->
342
+ ## Compatibility
343
+
344
+ The files provided are tested to work with:
345
+
346
+ - [text-generation-webui](https://github.com/oobabooga/text-generation-webui) using `Loader: AutoAWQ`.
347
+ - [vLLM](https://github.com/vllm-project/vllm) version 0.2.0 and later.
348
+ - [Hugging Face Text Generation Inference (TGI)](https://github.com/huggingface/text-generation-inference) version 1.1.0 and later.
349
+ - [Transformers](https://huggingface.co/docs/transformers) version 4.35.0 and later.
350
+ - [AutoAWQ](https://github.com/casper-hansen/AutoAWQ) version 0.1.1 and later.
351
+
352
+ <!-- README_AWQ.md-compatibility end -->
353
+
354
+ <!-- footer start -->
355
+ <!-- 200823 -->
356
+ ## Discord
357
+
358
+ For further support, and discussions on these models and AI in general, join us at:
359
+
360
+ [TheBloke AI's Discord server](https://discord.gg/theblokeai)
361
+
362
+ ## Thanks, and how to contribute
363
+
364
+ Thanks to the [chirper.ai](https://chirper.ai) team!
365
+
366
+ Thanks to Clay from [gpus.llm-utils.org](llm-utils)!
367
+
368
+ I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
369
+
370
+ If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
371
+
372
+ Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
373
+
374
+ * Patreon: https://patreon.com/TheBlokeAI
375
+ * Ko-Fi: https://ko-fi.com/TheBlokeAI
376
+
377
+ **Special thanks to**: Aemon Algiz.
378
+
379
+ **Patreon special mentions**: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
380
+
381
+
382
+ Thank you to all my generous patrons and donaters!
383
+
384
+ And thank you again to a16z for their generous grant.
385
+
386
+ <!-- footer end -->
387
+
388
+ # Original model card: Feynman Innovations's Python Code 13B
389
+
390
+
391
+ **Python-Code-13B**
392
+
393
+ Large Language Models (LLMs) are good with code generations. Sometimes LLMs do make mistakes in code generation. How about if they can give detailed explanation along with the code.
394
+ This is what I have tried over here. The base Llama-2 model was used for training purpose. It is trained on around 23000+ set of codes. Each set having 2 conversations.
395
+ This data was generated using GPT-3.5, GPT-4 etc. This conversation is in Vicuna/ShareGPT format. Each set, along with code, has detailed explanation.
396
+ I have released the [data](https://huggingface.co/datasets/ajibawa-2023/Python-Code-23k-ShareGPT).
397
+
398
+ **Training:**
399
+ Entire dataset was trained on Azure 4 x A100 80GB. For 3 epoch, training took 13 hours. DeepSpeed codebase was used for training purpose. This was trained on Llama-2 by Meta.
400
+
401
+
402
+ **GPTQ GGML & AWQ**
403
+
404
+ GPTQ: TBA
405
+
406
+ GGUF: TBA
407
+
408
+ AWQ: TBA
409
+
410
+
411
+ **Example Prompt:**
412
+ ```
413
+ This is a conversation with your helpful AI assistant. AI assistant can generate Python Code along with necessary explanation.
414
+
415
+ Context
416
+ You are a helpful AI assistant.
417
+
418
+ USER: <prompt>
419
+ ASSISTANT:
420
+ ```