TheBloke commited on
Commit
8e75104
1 Parent(s): 1b608b2

Initial GGUF model commit

Browse files
Files changed (1) hide show
  1. README.md +83 -93
README.md CHANGED
@@ -1,14 +1,16 @@
1
  ---
2
  inference: false
 
 
3
  license: llama2
4
  model_creator: Meta
5
- model_link: https://ai.meta.com/resources/models-and-libraries/llama-downloads
6
  model_name: CodeLlama 34B
7
  model_type: llama
 
8
  quantized_by: TheBloke
9
  tags:
10
  - llama-2
11
- - codellama
12
  ---
13
 
14
  <!-- header start -->
@@ -30,11 +32,11 @@ tags:
30
 
31
  # CodeLlama 34B - GGUF
32
  - Model creator: [Meta](https://huggingface.co/meta-llama)
33
- - Original model: [CodeLlama 34B](https://ai.meta.com/resources/models-and-libraries/llama-downloads)
34
 
35
  ## Description
36
 
37
- This repo contains GGUF format model files for [Meta's CodeLlama 34B](https://ai.meta.com/resources/models-and-libraries/llama-downloads).
38
 
39
  <!-- README_GGUF.md-about-gguf start -->
40
  ### About GGUF
@@ -45,15 +47,15 @@ The key benefit of GGUF is that it is a extensible, future-proof format which st
45
 
46
  As of August 25th, here is a list of clients and libraries that are known to support GGUF:
47
  * [llama.cpp](https://github.com/ggerganov/llama.cpp)
 
48
  * [KoboldCpp](https://github.com/LostRuins/koboldcpp), now supports GGUF as of release 1.41! A powerful GGML web UI, with full GPU accel. Especially good for story telling.
49
  * [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), should now work, choose the `c_transformers` backend. A great web UI with many interesting features. Supports CUDA GPU acceleration.
50
  * [ctransformers](https://github.com/marella/ctransformers), now supports GGUF as of version 0.2.24! A Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
 
51
  * [candle](https://github.com/huggingface/candle), added GGUF support on August 22nd. Candle is a Rust ML framework with a focus on performance, including GPU support, and ease of use.
52
 
53
- The clients and libraries below are expecting to add GGUF support. Where possible a link to the relevant issue or PR is provided:
54
- * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), awaiting llama-cpp-python support.
55
- * [LM Studio](https://lmstudio.ai/), in active development - hoped to be ready by August 25th-26th.
56
- * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), [in active development](https://github.com/abetlen/llama-cpp-python/issues/628).
57
  <!-- README_GGUF.md-about-gguf end -->
58
 
59
  <!-- repositories-available start -->
@@ -62,7 +64,7 @@ The clients and libraries below are expecting to add GGUF support. Where possibl
62
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/CodeLlama-34B-GPTQ)
63
  * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/CodeLlama-34B-GGUF)
64
  * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/CodeLlama-34B-GGML)
65
- * [Meta's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/CodeLlama-34B-fp16)
66
  <!-- repositories-available end -->
67
 
68
  <!-- prompt-template start -->
@@ -164,7 +166,7 @@ Donaters will get priority support on any and all AI/LLM/model questions and req
164
 
165
  **Special thanks to**: Aemon Algiz.
166
 
167
- **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
168
 
169
 
170
  Thank you to all my generous patrons and donaters!
@@ -176,125 +178,113 @@ And thank you again to a16z for their generous grant.
176
  <!-- original-model-card start -->
177
  # Original model card: Meta's CodeLlama 34B
178
 
 
 
179
 
180
- <!-- header start -->
181
- <!-- 200823 -->
182
- <div style="width: auto; margin-left: auto; margin-right: auto">
183
- <img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
184
- </div>
185
- <div style="display: flex; justify-content: space-between; width: 100%;">
186
- <div style="display: flex; flex-direction: column; align-items: flex-start;">
187
- <p style="margin-top: 0.5em; margin-bottom: 0em;"><a href="https://discord.gg/theblokeai">Chat & support: TheBloke's Discord server</a></p>
188
- </div>
189
- <div style="display: flex; flex-direction: column; align-items: flex-end;">
190
- <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>
191
- </div>
192
- </div>
193
- <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>
194
- <hr style="margin-top: 1.0em; margin-bottom: 1.0em;">
195
- <!-- header end -->
196
 
197
- # CodeLlama 34B fp16
198
- - Model creator: [Meta](https://ai.meta.com/llama/)
199
-
200
- ## Description
201
-
202
- This is Transformers/HF format fp16 weights for CodeLlama 34B. It is the result of downloading CodeLlama 34B from [Meta](https://ai.meta.com/blog/code-llama-large-language-model-coding/) and converting to HF using `convert_llama_weights_to_hf.py`.
203
-
204
- Quantisations will be coming shortly.
205
-
206
- Please note that due to a change in the RoPE Theta value, for correct results you must load these FP16 models with `trust_remote_code=True`
207
-
208
- Credit to @emozilla for creating the necessary modelling code to achieve this!
209
-
210
- ## Prompt template: TBC
211
 
 
212
 
213
- <!-- footer start -->
214
- <!-- 200823 -->
215
- ## Discord
216
-
217
- For further support, and discussions on these models and AI in general, join us at:
218
-
219
- [TheBloke AI's Discord server](https://discord.gg/theblokeai)
220
-
221
- ## Thanks, and how to contribute.
222
-
223
- Thanks to the [chirper.ai](https://chirper.ai) team!
224
-
225
- 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.
226
-
227
- 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.
228
-
229
- Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
230
-
231
- * Patreon: https://patreon.com/TheBlokeAI
232
- * Ko-Fi: https://ko-fi.com/TheBlokeAI
233
-
234
- **Special thanks to**: Aemon Algiz.
235
-
236
- **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
237
-
238
-
239
- Thank you to all my generous patrons and donaters!
240
 
241
- And thank you again to a16z for their generous grant.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
242
 
243
- <!-- footer end -->
244
 
245
- # Original model card
 
246
 
247
- # Code Llama
248
 
249
- ## **Model Details**
250
 
251
- **Model Developers** Meta AI
 
 
252
 
253
- **Variations** Code Llama comes in three model sizes, and three variants:
254
- 1) Code Llama: our base models designed for general code synthesis and understanding
255
- 2) Code Llama - Python: designed specifically for Python
256
- 3) Code Llama - Instruct: for instruction following and safer deployment
257
-
258
  All variants are available in sizes of 7B, 13B and 34B parameters.
259
 
 
 
260
  **Input** Models input text only.
261
 
262
- **Output** Models output text only.
263
 
264
- **Model Architecture** Code Llama and its variants are autoregressive language models using optimized transformer architectures. Code Llama 7B and 13B additionally support infilling text generation. All models were fine-tuned with up to 16K tokens, and support up to 100K tokens at inference time.
265
 
266
  **Model Dates** Code Llama and its variants have been trained between January 2023 and July 2023.
267
 
268
- **Status** This is a static model trained on an offline dataset. Future versions of Code Llama - Instruct will be released as we improve model safety with community feedback.
269
 
270
- **Licence** A custom commercial license is available at: [https://ai.meta.com/resources/models-and-libraries/llama-downloads/](https://ai.meta.com/resources/models-and-libraries/llama-downloads/).
271
 
272
  **Research Paper** More information can be found in the paper "[Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)".
273
 
274
- **Where to send comments** Instructions on how to provide feedback or comments on the model can be found in the model [README](README.md), or by opening an issue in the GitHub repository ([https://github.com/facebookresearch/codellama/](https://github.com/facebookresearch/codellama/)).
275
-
276
- ## **Intended Use**
277
  **Intended Use Cases** Code Llama and its variants is intended for commercial and research use in English and relevant programming languages. The base model Code Llama can be adapted for a variety of code synthesis and understanding tasks, Code Llama - Python is designed specifically to handle the Python programming language, and Code Llama - Instruct is intended to be safer to use for code assistant and generation applications.
278
 
279
  **Out-of-Scope Uses** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. Use in any other way that is prohibited by the Acceptable Use Policy and Licensing Agreement for Code Llama and its variants.
280
 
281
- ## **Hardware and Software**
282
- **Training Factors**
283
- We used custom training libraries. The training and fine-tuning of the released models have been performed Meta’s Research Super Cluster.
284
 
285
  **Carbon Footprint** In aggregate, training all 9 Code Llama models required 400K GPU hours of computation on hardware of type A100-80GB (TDP of 350-400W). Estimated total emissions were 65.3 tCO2eq, 100% of which were offset by Meta’s sustainability program.
286
 
287
- **Training data**
 
288
  All experiments reported here and the released models have been trained and fine-tuned using the same data as Llama 2 with different weights (see Section 2 and Table 1 in the [research paper](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) for details).
289
- Code Llama - Instruct uses additional instruction fine-tuning data.
290
 
291
- **Evaluation Results**
 
292
  See evaluations for the main models and detailed ablations in Section 3 and safety evaluations in Section 4 of the research paper.
293
 
294
- ## **Ethical Considerations and Limitations**
 
 
295
  Code Llama and its variants are a new technology that carries risks with use. Testing conducted to date has been in English, and has not covered, nor could it cover all scenarios. For these reasons, as with all LLMs, Code Llama’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate or objectionable responses to user prompts. Therefore, before deploying any applications of Code Llama, developers should perform safety testing and tuning tailored to their specific applications of the model.
296
 
297
  Please see the Responsible Use Guide available available at [https://ai.meta.com/llama/responsible-user-guide](https://ai.meta.com/llama/responsible-user-guide).
298
 
299
-
300
  <!-- original-model-card end -->
 
1
  ---
2
  inference: false
3
+ language:
4
+ - code
5
  license: llama2
6
  model_creator: Meta
7
+ model_link: https://huggingface.co/codellama/CodeLlama-34b-hf
8
  model_name: CodeLlama 34B
9
  model_type: llama
10
+ pipeline_tag: text-generation
11
  quantized_by: TheBloke
12
  tags:
13
  - llama-2
 
14
  ---
15
 
16
  <!-- header start -->
 
32
 
33
  # CodeLlama 34B - GGUF
34
  - Model creator: [Meta](https://huggingface.co/meta-llama)
35
+ - Original model: [CodeLlama 34B](https://huggingface.co/codellama/CodeLlama-34b-hf)
36
 
37
  ## Description
38
 
39
+ This repo contains GGUF format model files for [Meta's CodeLlama 34B](https://huggingface.co/codellama/CodeLlama-34b-hf).
40
 
41
  <!-- README_GGUF.md-about-gguf start -->
42
  ### About GGUF
 
47
 
48
  As of August 25th, here is a list of clients and libraries that are known to support GGUF:
49
  * [llama.cpp](https://github.com/ggerganov/llama.cpp)
50
+ * [text-generation-webui](https://github.com/oobabooga/text-generation-webui), the most widely used web UI. Supports GGUF with GPU acceleration via the ctransformers backend - llama-cpp-python backend should work soon too.
51
  * [KoboldCpp](https://github.com/LostRuins/koboldcpp), now supports GGUF as of release 1.41! A powerful GGML web UI, with full GPU accel. Especially good for story telling.
52
  * [LoLLMS Web UI](https://github.com/ParisNeo/lollms-webui), should now work, choose the `c_transformers` backend. A great web UI with many interesting features. Supports CUDA GPU acceleration.
53
  * [ctransformers](https://github.com/marella/ctransformers), now supports GGUF as of version 0.2.24! A Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
54
+ * [llama-cpp-python](https://github.com/abetlen/llama-cpp-python), supports GGUF as of version 0.1.79. A Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
55
  * [candle](https://github.com/huggingface/candle), added GGUF support on August 22nd. Candle is a Rust ML framework with a focus on performance, including GPU support, and ease of use.
56
 
57
+ The clients and libraries below are expecting to add GGUF support shortly:
58
+ * [LM Studio](https://lmstudio.ai/), should be updated by end August 25th.
 
 
59
  <!-- README_GGUF.md-about-gguf end -->
60
 
61
  <!-- repositories-available start -->
 
64
  * [GPTQ models for GPU inference, with multiple quantisation parameter options.](https://huggingface.co/TheBloke/CodeLlama-34B-GPTQ)
65
  * [2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference](https://huggingface.co/TheBloke/CodeLlama-34B-GGUF)
66
  * [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference (deprecated)](https://huggingface.co/TheBloke/CodeLlama-34B-GGML)
67
+ * [Meta's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/codellama/CodeLlama-34b-hf)
68
  <!-- repositories-available end -->
69
 
70
  <!-- prompt-template start -->
 
166
 
167
  **Special thanks to**: Aemon Algiz.
168
 
169
+ **Patreon special mentions**: Kacper Wikieł, knownsqashed, Leonard Tan, Asp the Wyvern, Daniel P. Andersen, Luke Pendergrass, Stanislav Ovsiannikov, RoA, Dave, Ai Maven, Kalila, Will Dee, Imad Khwaja, Nitin Borwankar, Joseph William Delisle, Tony Hughes, Cory Kujawski, Rishabh Srivastava, Russ Johnson, Stephen Murray, Lone Striker, Johann-Peter Hartmann, Elle, J, Deep Realms, SuperWojo, Raven Klaugh, Sebastain Graf, ReadyPlayerEmma, Alps Aficionado, Mano Prime, Derek Yates, Gabriel Puliatti, Mesiah Bishop, Magnesian, Sean Connelly, biorpg, Iucharbius, Olakabola, Fen Risland, Space Cruiser, theTransient, Illia Dulskyi, Thomas Belote, Spencer Kim, Pieter, John Detwiler, Fred von Graf, Michael Davis, Swaroop Kallakuri, subjectnull, Clay Pascal, Subspace Studios, Chris Smitley, Enrico Ros, usrbinkat, Steven Wood, alfie_i, David Ziegler, Willem Michiel, Matthew Berman, Andrey, Pyrater, Jeffrey Morgan, vamX, LangChain4j, Luke @flexchar, Trenton Dambrowitz, Pierre Kircher, Alex, Sam, James Bentley, Edmond Seymore, Eugene Pentland, Pedro Madruga, Rainer Wilmers, Dan Guido, Nathan LeClaire, Spiking Neurons AB, Talal Aujan, zynix, Artur Olbinski, Michael Levine, 阿明, K, John Villwock, Nikolai Manek, Femi Adebogun, senxiiz, Deo Leter, NimbleBox.ai, Viktor Bowallius, Geoffrey Montalvo, Mandus, Ajan Kanaga, ya boyyy, Jonathan Leane, webtim, Brandon Frisco, danny, Alexandros Triantafyllidis, Gabriel Tamborski, Randy H, terasurfer, Vadim, Junyu Yang, Vitor Caleffi, Chadd, transmissions 11
170
 
171
 
172
  Thank you to all my generous patrons and donaters!
 
178
  <!-- original-model-card start -->
179
  # Original model card: Meta's CodeLlama 34B
180
 
181
+ # **Code Llama**
182
+ Code Llama is a collection of pretrained and fine-tuned generative text models ranging in scale from 7 billion to 34 billion parameters. This is the repository for the base 34B version in the Hugging Face Transformers format. This model is designed for general code synthesis and understanding. Links to other models can be found in the index at the bottom.
183
 
184
+ | | Base Model | Python | Instruct |
185
+ | --- | ----------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
186
+ | 7B | [codellama/CodeLlama-7b-hf](https://huggingface.co/codellama/CodeLlama-7b-hf) | [codellama/CodeLlama-7b-Python-hf](https://huggingface.co/codellama/CodeLlama-7b-Python-hf) | [codellama/CodeLlama-7b-Instruct-hf](https://huggingface.co/codellama/CodeLlama-7b-Instruct-hf) |
187
+ | 13B | [codellama/CodeLlama-13b-hf](https://huggingface.co/codellama/CodeLlama-13b-hf) | [codellama/CodeLlama-13b-Python-hf](https://huggingface.co/codellama/CodeLlama-13b-Python-hf) | [codellama/CodeLlama-13b-Instruct-hf](https://huggingface.co/codellama/CodeLlama-13b-Instruct-hf) |
188
+ | 34B | [codellama/CodeLlama-34b-hf](https://huggingface.co/codellama/CodeLlama-34b-hf) | [codellama/CodeLlama-34b-Python-hf](https://huggingface.co/codellama/CodeLlama-34b-Python-hf) | [codellama/CodeLlama-34b-Instruct-hf](https://huggingface.co/codellama/CodeLlama-34b-Instruct-hf) |
 
 
 
 
 
 
 
 
 
 
 
189
 
190
+ ## Model Use
 
 
 
 
 
 
 
 
 
 
 
 
 
191
 
192
+ To use this model, please make sure to install transformers from `main` until the next version is released:
193
 
194
+ ```bash
195
+ pip install git+https://github.com/huggingface/transformers.git@main accelerate
196
+ ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
197
 
198
+ Model capabilities:
199
+
200
+ - [x] Code completion.
201
+ - [ ] Infilling.
202
+ - [ ] Instructions / chat.
203
+ - [ ] Python specialist.
204
+
205
+
206
+ ```python
207
+ from transformers import AutoTokenizer
208
+ import transformers
209
+ import torch
210
+
211
+ model = "codellama/CodeLlama-34b-hf"
212
+
213
+ tokenizer = AutoTokenizer.from_pretrained(model)
214
+ pipeline = transformers.pipeline(
215
+ "text-generation",
216
+ model=model,
217
+ torch_dtype=torch.float16,
218
+ device_map="auto",
219
+ )
220
+
221
+ sequences = pipeline(
222
+ 'import socket\n\ndef ping_exponential_backoff(host: str):',
223
+ do_sample=True,
224
+ top_k=10,
225
+ temperature=0.1,
226
+ top_p=0.95,
227
+ num_return_sequences=1,
228
+ eos_token_id=tokenizer.eos_token_id,
229
+ max_length=200,
230
+ )
231
+ for seq in sequences:
232
+ print(f"Result: {seq['generated_text']}")
233
+ ```
234
 
 
235
 
236
+ ## Model Details
237
+ *Note: Use of this model is governed by the Meta license. Meta developed and publicly released the Code Llama family of large language models (LLMs).
238
 
239
+ **Model Developers** Meta
240
 
241
+ **Variations** Code Llama comes in three model sizes, and three variants:
242
 
243
+ * Code Llama: base models designed for general code synthesis and understanding
244
+ * Code Llama - Python: designed specifically for Python
245
+ * Code Llama - Instruct: for instruction following and safer deployment
246
 
 
 
 
 
 
247
  All variants are available in sizes of 7B, 13B and 34B parameters.
248
 
249
+ **This repository contains the base version of the 34B parameters model.**
250
+
251
  **Input** Models input text only.
252
 
253
+ **Output** Models generate text only.
254
 
255
+ **Model Architecture** Code Llama is an auto-regressive language model that uses an optimized transformer architecture.
256
 
257
  **Model Dates** Code Llama and its variants have been trained between January 2023 and July 2023.
258
 
259
+ **Status** This is a static model trained on an offline dataset. Future versions of Code Llama - Instruct will be released as we improve model safety with community feedback.
260
 
261
+ **License** A custom commercial license is available at: [https://ai.meta.com/resources/models-and-libraries/llama-downloads/](https://ai.meta.com/resources/models-and-libraries/llama-downloads/)
262
 
263
  **Research Paper** More information can be found in the paper "[Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)".
264
 
265
+ ## Intended Use
 
 
266
  **Intended Use Cases** Code Llama and its variants is intended for commercial and research use in English and relevant programming languages. The base model Code Llama can be adapted for a variety of code synthesis and understanding tasks, Code Llama - Python is designed specifically to handle the Python programming language, and Code Llama - Instruct is intended to be safer to use for code assistant and generation applications.
267
 
268
  **Out-of-Scope Uses** Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in languages other than English. Use in any other way that is prohibited by the Acceptable Use Policy and Licensing Agreement for Code Llama and its variants.
269
 
270
+ ## Hardware and Software
271
+ **Training Factors** We used custom training libraries. The training and fine-tuning of the released models have been performed Meta’s Research Super Cluster.
 
272
 
273
  **Carbon Footprint** In aggregate, training all 9 Code Llama models required 400K GPU hours of computation on hardware of type A100-80GB (TDP of 350-400W). Estimated total emissions were 65.3 tCO2eq, 100% of which were offset by Meta’s sustainability program.
274
 
275
+ ## Training Data
276
+
277
  All experiments reported here and the released models have been trained and fine-tuned using the same data as Llama 2 with different weights (see Section 2 and Table 1 in the [research paper](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) for details).
 
278
 
279
+ ## Evaluation Results
280
+
281
  See evaluations for the main models and detailed ablations in Section 3 and safety evaluations in Section 4 of the research paper.
282
 
283
+
284
+ ## Ethical Considerations and Limitations
285
+
286
  Code Llama and its variants are a new technology that carries risks with use. Testing conducted to date has been in English, and has not covered, nor could it cover all scenarios. For these reasons, as with all LLMs, Code Llama’s potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate or objectionable responses to user prompts. Therefore, before deploying any applications of Code Llama, developers should perform safety testing and tuning tailored to their specific applications of the model.
287
 
288
  Please see the Responsible Use Guide available available at [https://ai.meta.com/llama/responsible-user-guide](https://ai.meta.com/llama/responsible-user-guide).
289
 
 
290
  <!-- original-model-card end -->