From 614f525bd2099947139f219ddd423b13272c7764 Mon Sep 17 00:00:00 2001 From: DejianYang Date: Fri, 3 Nov 2023 11:44:05 +0800 Subject: [PATCH 01/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 9bddcdf..5256286 100644 --- a/README.md +++ b/README.md @@ -257,7 +257,7 @@ The reproducible code for the following evaluation results can be found in the [ ![Math](pictures/Math.png) -### 6. Lincense +### 6. Lincese This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the Model License. DeepSeek Coder supports commercial use. See the [LICENSE-CODE](LICENSE-CODE) and [LICENSE-MODEL](LICENSE-MODEL) for more details. From 64817a3179348bcd817728950ed800b0e6f62197 Mon Sep 17 00:00:00 2001 From: DejianYang Date: Fri, 3 Nov 2023 13:18:17 +0800 Subject: [PATCH 02/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 5256286..3dde1b6 100644 --- a/README.md +++ b/README.md @@ -257,7 +257,7 @@ The reproducible code for the following evaluation results can be found in the [ ![Math](pictures/Math.png) -### 6. Lincese +### 6. License This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the Model License. DeepSeek Coder supports commercial use. See the [LICENSE-CODE](LICENSE-CODE) and [LICENSE-MODEL](LICENSE-MODEL) for more details. From 1a328e9e5e42f0ee8834aaec9d8f24e35f8db43a Mon Sep 17 00:00:00 2001 From: ZHU QIHAO <18811325956@163.com> Date: Fri, 3 Nov 2023 20:58:44 +0800 Subject: [PATCH 03/35] Create requirements.txt --- requirements.txt | 8 ++++++++ 1 file changed, 8 insertions(+) create mode 100644 requirements.txt diff --git a/requirements.txt b/requirements.txt new file mode 100644 index 0000000..fb2486a --- /dev/null +++ b/requirements.txt @@ -0,0 +1,8 @@ +torch>=2.0 +tokenizers>=0.14.0 +transformers>=4..35.0 +accelerate +sympy==1.12 +pebble +timeout-decorator +attrdict From a587295ba50086553d7509d6710f481b280c5edd Mon Sep 17 00:00:00 2001 From: ZHU QIHAO <18811325956@163.com> Date: Fri, 3 Nov 2023 21:01:52 +0800 Subject: [PATCH 04/35] Update README.md --- README.md | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 3dde1b6..1fe7828 100644 --- a/README.md +++ b/README.md @@ -5,7 +5,7 @@
-### 1. Introduction of Deepseek Coder +### 1. Introduction of DeepSeek Coder Deepseek Coder comprises a series of code language models trained on both 87% code and 13% natural language in English and Chinese, with each model pre-trained on 2T tokens. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks. @@ -62,7 +62,11 @@ More evaluation details can be found in the [Detailed Evaluation](#5-detailed-ev ### 4. How to Use -Here give some examples of how to use our model. +Before proceeding, you'll need to install the necessary dependencies. You can do this by running the following command: +``` +pip install -r requirements.txt +``` +Here are some examples of how to use our model. #### 1）Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM From 405152b0709330f29e3f0d5569e07157768328cc Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 22:31:24 +0800 Subject: [PATCH 05/35] Update README.md --- README.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/README.md b/README.md index 1fe7828..19e4d9b 100644 --- a/README.md +++ b/README.md @@ -67,6 +67,8 @@ Before proceeding, you'll need to install the necessary dependencies. You can do pip install -r requirements.txt ``` Here are some examples of how to use our model. + +A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct) (thanks to all the HF team for their support). #### 1）Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM From 44cf16ebda295fcbfc11512b0fb10e618a4ec4fc Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 22:41:45 +0800 Subject: [PATCH 06/35] Update README.md --- README.md | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 19e4d9b..fc949da 100644 --- a/README.md +++ b/README.md @@ -66,9 +66,10 @@ Before proceeding, you'll need to install the necessary dependencies. You can do ``` pip install -r requirements.txt ``` +A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct), and you can run the demo locally using [app.py](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct/blob/main/app.py). (thanks to all the HF team for their support) + Here are some examples of how to use our model. -A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct) (thanks to all the HF team for their support). #### 1）Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM From 299d79263c49f71e09cf8937c1a2c97a03d5fa68 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 23:06:40 +0800 Subject: [PATCH 07/35] Create app.py --- demo/app.py | 129 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 129 insertions(+) create mode 100644 demo/app.py diff --git a/demo/app.py b/demo/app.py new file mode 100644 index 0000000..dec892f --- /dev/null +++ b/demo/app.py @@ -0,0 +1,129 @@ +import os +from threading import Thread +from typing import Iterator + +import gradio as gr +import spaces +import torch +from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer + +MAX_MAX_NEW_TOKENS = 2048 +DEFAULT_MAX_NEW_TOKENS = 1024 + +MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096")) + +DESCRIPTION = """\ +# DeepSeek-6.7B-Chat + +This Space demonstrates model [DeepSeek-Coder](https://huggingface.co/deepseek-ai/deepseek-coder-6.7b-instruct) by DeepSeek, a code model with 6.7B parameters fine-tuned for chat instructions. +""" + +if not torch.cuda.is_available(): + DESCRIPTION += "\n

Running on CPU 🥶 This demo does not work on CPU.

" + + +if torch.cuda.is_available(): + model_id = "deepseek-ai/deepseek-coder-6.7b-instruct" + model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto") + tokenizer = AutoTokenizer.from_pretrained(model_id) + tokenizer.use_default_system_prompt = False + + + +@spaces.GPU +def generate( + message: str, + chat_history: list, + system_prompt: str, + max_new_tokens: int = 1024, + temperature: float = 0.6, + top_p: float = 0.9, + top_k: int = 50, + repetition_penalty: float = 1, +) -> Iterator[str]: + conversation = [] + if system_prompt: + conversation.append({"role": "system", "content": system_prompt}) + for user, assistant in chat_history: + conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}]) + conversation.append({"role": "user", "content": message}) + + input_ids = tokenizer.apply_chat_template(conversation, return_tensors="pt") + if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH: + input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:] + gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.") + input_ids = input_ids.to(model.device) + + streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True) + generate_kwargs = dict( + {"input_ids": input_ids}, + streamer=streamer, + max_new_tokens=max_new_tokens, + do_sample=False, + num_beams=1, + repetition_penalty=repetition_penalty, + eos_token_id=32021 + ) + t = Thread(target=model.generate, kwargs=generate_kwargs) + t.start() + + outputs = [] + for text in streamer: + outputs.append(text) + yield "".join(outputs).replace("<|EOT|>","") + + +chat_interface = gr.ChatInterface( + fn=generate, + additional_inputs=[ + gr.Textbox(label="System prompt", lines=6), + gr.Slider( + label="Max new tokens", + minimum=1, + maximum=MAX_MAX_NEW_TOKENS, + step=1, + value=DEFAULT_MAX_NEW_TOKENS, + ), + # gr.Slider( + # label="Temperature", + # minimum=0, + # maximum=4.0, + # step=0.1, + # value=0, + # ), + gr.Slider( + label="Top-p (nucleus sampling)", + minimum=0.05, + maximum=1.0, + step=0.05, + value=0.9, + ), + gr.Slider( + label="Top-k", + minimum=1, + maximum=1000, + step=1, + value=50, + ), + gr.Slider( + label="Repetition penalty", + minimum=1.0, + maximum=2.0, + step=0.05, + value=1, + ), + ], + stop_btn=None, + examples=[ + ["implement snake game using pygame"], + ["Can you explain briefly to me what is the Python programming language?"], + ["write a program to find the factorial of a number"], + ], +) + +with gr.Blocks(css="style.css") as demo: + gr.Markdown(DESCRIPTION) + chat_interface.render() + +if __name__ == "__main__": + demo.queue().launch(share=True) From 4d3115d7f8b780142399dd4244987690d23ddf01 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 23:07:15 +0800 Subject: [PATCH 08/35] Create requirement.txt --- demo/requirement.txt | 9 +++++++++ 1 file changed, 9 insertions(+) create mode 100644 demo/requirement.txt diff --git a/demo/requirement.txt b/demo/requirement.txt new file mode 100644 index 0000000..640a5d3 --- /dev/null +++ b/demo/requirement.txt @@ -0,0 +1,9 @@ +accelerate==0.23.0 +bitsandbytes==0.41.1 +gradio==3.48.0 +protobuf==3.20.3 +scipy==1.10.1 +sentencepiece==0.1.99 +spaces==0.16.1 +torch==2.0.0 +transformers==4.34.0 From 7b96cb119da79344463b6e3626983c04107b6c53 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 23:07:31 +0800 Subject: [PATCH 09/35] Create style.css --- demo/style.css | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) create mode 100644 demo/style.css diff --git a/demo/style.css b/demo/style.css new file mode 100644 index 0000000..60878fe --- /dev/null +++ b/demo/style.css @@ -0,0 +1,16 @@ +h1 { + text-align: center; +} + +#duplicate-button { + margin: auto; + color: white; + background: #1565c0; + border-radius: 100vh; +} + +.contain { + max-width: 900px; + margin: auto; + padding-top: 1.5rem; +} From 38991b90381af2057e3b0753075674cf4c355e41 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 23:09:23 +0800 Subject: [PATCH 10/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index fc949da..b7c4e95 100644 --- a/README.md +++ b/README.md @@ -66,7 +66,7 @@ Before proceeding, you'll need to install the necessary dependencies. You can do ``` pip install -r requirements.txt ``` -A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct), and you can run the demo locally using [app.py](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct/blob/main/app.py). (thanks to all the HF team for their support) +A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct), and you can run the demo locally using `app.py` in [demo](https://github.com/deepseek-ai/deepseek-coder/tree/main/demo) folder. (Thanks to all the HF team for their support) Here are some examples of how to use our model. From 1af85c9ad616a6b9c33d812f6db90409953c7bc6 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Fri, 3 Nov 2023 23:43:08 +0800 Subject: [PATCH 11/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index b7c4e95..372c0a6 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [Wechat(微信)]

From f3d62df86840667d8798a425e9df4f283418edb9 Mon Sep 17 00:00:00 2001 From: Benjamin-eecs Date: Sat, 4 Nov 2023 10:32:08 +0800 Subject: [PATCH 12/35] fix: update discord invitation link --- README.md | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index 372c0a6..6d0bce2 100644 --- a/README.md +++ b/README.md @@ -1,24 +1,24 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [Wechat(微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [Wechat(微信)]

### 1. Introduction of DeepSeek Coder -Deepseek Coder comprises a series of code language models trained on both 87% code and 13% natural language in English and Chinese, with each model pre-trained on 2T tokens. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks. +Deepseek Coder comprises a series of code language models trained on both 87% code and 13% natural language in English and Chinese, with each model pre-trained on 2T tokens. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks.

result

- **Massive Training Data**: Trained on 2T tokens, including 87% code and 13% linguistic data in both English and Chinese languages. - + - **Highly Flexible & Scalable**: Offered in model sizes of 1B, 5.7B, 6.7B and 33B, enabling users to choose the setup most suitable for their requirements. - + - **Superior Model Performance**: State-of-the-art performance among publicly available code models on HumanEval, MultiPL-E, MBPP, DS-1000, and APPS benchmarks. - + - **Advanced Code Completion Capabilities**: A window size of 16K and a fill-in-the-blank task, supporting project-level code completion and infilling tasks. @@ -187,7 +187,7 @@ def load_data(): X_test = torch.tensor(X_test, dtype=torch.float32) y_train = torch.tensor(y_train, dtype=torch.int64) y_test = torch.tensor(y_test, dtype=torch.int64) - + return X_train, X_test, y_train, y_test def evaluate_predictions(y_test, y_pred): @@ -213,7 +213,7 @@ class IrisClassifier(nn.Module): def train_model(self, X_train, y_train, epochs, lr, batch_size): criterion = nn.CrossEntropyLoss() optimizer = optim.Adam(self.parameters(), lr=lr) - + # Create DataLoader for batches dataset = TensorDataset(X_train, y_train) dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=True) @@ -272,4 +272,3 @@ See the [LICENSE-CODE](LICENSE-CODE) and [LICENSE-MODEL](LICENSE-MODEL) for more ### 6. Contact If you have any questions, please raise an issue or contact us at [agi_code@deepseek.com](mailto:agi_code@deepseek.com). - From a4e9ff390dcfe1dd6fb479da4d4da5ee5d437271 Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Sat, 4 Nov 2023 13:45:26 +0800 Subject: [PATCH 13/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 6d0bce2..d60e37d 100644 --- a/README.md +++ b/README.md @@ -66,7 +66,7 @@ Before proceeding, you'll need to install the necessary dependencies. You can do ``` pip install -r requirements.txt ``` -A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-7b-instruct), and you can run the demo locally using `app.py` in [demo](https://github.com/deepseek-ai/deepseek-coder/tree/main/demo) folder. (Thanks to all the HF team for their support) +A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-33b-instruct), and you can run the demo locally using `app.py` in [demo](https://github.com/deepseek-ai/deepseek-coder/tree/main/demo) folder. (Thanks to all the HF team for their support) Here are some examples of how to use our model. From 8b4290d068651ee27b4d1c3492ba4236bf0da587 Mon Sep 17 00:00:00 2001 From: ZHU QIHAO <18811325956@163.com> Date: Sat, 4 Nov 2023 14:34:55 +0800 Subject: [PATCH 14/35] Update README.md --- README.md | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/README.md b/README.md index d60e37d..06a850c 100644 --- a/README.md +++ b/README.md @@ -158,6 +158,20 @@ print("Sorted array:", quick_sort(arr)) This code works by selecting a 'pivot' element from the array and partitioning the other elements into two sub-arrays, according to whether they are less than or greater than the pivot. The pivot element is then in its final position. The process is then repeated for the sub-arrays. ``` +If you don't want to use the provided api `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. +``` +You are an AI programming assistant, utilizing the Deepseek Coder model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer. +### Instruction: +['content'] +### Response: +['content'] +<|EOT|> +### Instruction: +['content'] +### Response: + +``` + #### 4）Repository Level Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM From 077611e86b7a9df77dee9e553472b8cba11b5a6b Mon Sep 17 00:00:00 2001 From: ZHU QIHAO <18811325956@163.com> Date: Sat, 4 Nov 2023 14:38:48 +0800 Subject: [PATCH 15/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 06a850c..f03b982 100644 --- a/README.md +++ b/README.md @@ -158,7 +158,7 @@ print("Sorted array:", quick_sort(arr)) This code works by selecting a 'pivot' element from the array and partitioning the other elements into two sub-arrays, according to whether they are less than or greater than the pivot. The pivot element is then in its final position. The process is then repeated for the sub-arrays. ``` -If you don't want to use the provided api `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. +If you don't want to use the provided api `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. Replace the `['content']` with your instructions and the model's previous (if any) responses, then the model will generate the response to the currently given instruction. ``` You are an AI programming assistant, utilizing the Deepseek Coder model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer. ### Instruction: From 7ea6c3d87f12519d9f92ccf5011eedee3a1c9b7c Mon Sep 17 00:00:00 2001 From: ZHU QIHAO <18811325956@163.com> Date: Sun, 5 Nov 2023 11:47:41 +0800 Subject: [PATCH 16/35] Update requirements.txt --- requirements.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/requirements.txt b/requirements.txt index fb2486a..7b138c4 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,6 +1,6 @@ torch>=2.0 tokenizers>=0.14.0 -transformers>=4..35.0 +transformers>=4.35.0 accelerate sympy==1.12 pebble From 4f1e96627421c25d3320af826da2efdd7f5c652c Mon Sep 17 00:00:00 2001 From: Daya Guo <40300434+guoday@users.noreply.github.com> Date: Sun, 5 Nov 2023 13:16:56 +0800 Subject: [PATCH 17/35] Update requirement.txt --- demo/requirement.txt | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/demo/requirement.txt b/demo/requirement.txt index 640a5d3..7148fe0 100644 --- a/demo/requirement.txt +++ b/demo/requirement.txt @@ -6,4 +6,4 @@ scipy==1.10.1 sentencepiece==0.1.99 spaces==0.16.1 torch==2.0.0 -transformers==4.34.0 +transformers>=4.35.0 From 4e864b7b3315f4fa967706753876bc86f0efa4a0 Mon Sep 17 00:00:00 2001 From: Fuli Luo Date: Sun, 5 Nov 2023 18:17:08 +0800 Subject: [PATCH 18/35] Update README.md --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index f03b982..f8ee8be 100644 --- a/README.md +++ b/README.md @@ -7,13 +7,13 @@ ### 1. Introduction of DeepSeek Coder -Deepseek Coder comprises a series of code language models trained on both 87% code and 13% natural language in English and Chinese, with each model pre-trained on 2T tokens. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and a extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks. +Deepseek Coder consists of a series of code language models, each trained from scratch on a dataset of 2 trillion tokens, composed of 87% code and 13% natural language, encompassing both English and Chinese. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and an extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks.

result

-- **Massive Training Data**: Trained on 2T tokens, including 87% code and 13% linguistic data in both English and Chinese languages. +- **Massive Training Data**: Trained from scratch on 2T tokens, including 87% code and 13% linguistic data in both English and Chinese languages. - **Highly Flexible & Scalable**: Offered in model sizes of 1B, 5.7B, 6.7B and 33B, enabling users to choose the setup most suitable for their requirements. From 3e14fa666f59c853573acf6e793ace5436e3e85a Mon Sep 17 00:00:00 2001 From: Fuli Luo Date: Sun, 5 Nov 2023 18:24:53 +0800 Subject: [PATCH 19/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index f8ee8be..378a6c4 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,7 @@ ### 1. Introduction of DeepSeek Coder -Deepseek Coder consists of a series of code language models, each trained from scratch on a dataset of 2 trillion tokens, composed of 87% code and 13% natural language, encompassing both English and Chinese. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and an extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks. +Deepseek Coder is composed of a series of code language models, each trained from scratch on 2T tokens, with a composition of 87% code and 13% natural language in both English and Chinese. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and an extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks.

result From cb1df1de2a08440699d646d1aff95cd4408b475e Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Sun, 5 Nov 2023 23:51:51 +0800 Subject: [PATCH 20/35] Update README.md --- README.md | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/README.md b/README.md index 378a6c4..2ea714f 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [Wechat(微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

@@ -38,7 +38,6 @@ And the DeepSeek-Coder-Instruct-33B model after instruction tuning outperforms G More evaluation details can be found in the [Detailed Evaluation](#5-detailed-evaluation-results). - ### 3. Procedure of Data Creation and Model Training #### Data Creation From b4253af1b1ed87b766fc0c57c978f966079dafac Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Sun, 5 Nov 2023 23:53:22 +0800 Subject: [PATCH 21/35] Update README.md --- README.md | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index 2ea714f..ee4993d 100644 --- a/README.md +++ b/README.md @@ -69,7 +69,7 @@ A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co Here are some examples of how to use our model. -#### 1）Code Completion +#### 1) Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM import torch @@ -96,7 +96,7 @@ def quick_sort(arr): return quick_sort(left) + [pivot] + quick_sort(right) ``` -#### 2）Code Insertion +#### 2) Code Insertion ```python from transformers import AutoTokenizer, AutoModelForCausalLM import torch @@ -123,7 +123,7 @@ This code will output the following result: for i in range(1, len(arr)): ``` -#### 3）Chat Model Inference +#### 3) Chat Model Inference ```python from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/deepseek-coder-6.7b-instruct", trust_remote_code=True) @@ -171,7 +171,7 @@ You are an AI programming assistant, utilizing the Deepseek Coder model, develop ``` -#### 4）Repository Level Code Completion +#### 4) Repository Level Code Completion ```python from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/deepseek-coder-6.7b-base", trust_remote_code=True) @@ -264,16 +264,16 @@ In the following scenario, the Deepseek-Coder 6.7B model effectively calls a cla ### 5. Detailed Evaluation Results The reproducible code for the following evaluation results can be found in the [Evaluation](https://github.com/deepseek-ai/deepseek-coder/tree/main/Evaluation) directory. -#### 1）Multilingual HumanEval Benchmark +#### 1) Multilingual HumanEval Benchmark ![HumanEval](pictures/HumanEval.png) -#### 2）MBPP Benchmark +#### 2) MBPP Benchmark MBPP -#### 3）DS-1000 Benchmark +#### 3) DS-1000 Benchmark ![DS-1000](pictures/DS-1000.png) -#### 4）Program-Aid Math Reasoning Benchmark +#### 4) Program-Aid Math Reasoning Benchmark ![Math](pictures/Math.png) From 6eb5822ca71050dfc93d2350c1fc83d0743da97a Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Sun, 5 Nov 2023 23:55:22 +0800 Subject: [PATCH 22/35] Update README.md --- README.md | 2 -- 1 file changed, 2 deletions(-) diff --git a/README.md b/README.md index ee4993d..433d9f2 100644 --- a/README.md +++ b/README.md @@ -58,8 +58,6 @@ More evaluation details can be found in the [Detailed Evaluation](#5-detailed-ev model_pretraining - - ### 4. How to Use Before proceeding, you'll need to install the necessary dependencies. You can do this by running the following command: ``` From 96aef33e8c87e76f321e2d59e94ad14cce3f72ca Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Sun, 5 Nov 2023 23:57:24 +0800 Subject: [PATCH 23/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 433d9f2..0d0c42c 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

From 89655af972cc53f7e20feb13981e260dfbc1bba8 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Sun, 5 Nov 2023 23:57:59 +0800 Subject: [PATCH 24/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 0d0c42c..96c68a2 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

From 0f4739607f696c39202fb01ee29e865cacadf234 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:01:05 +0800 Subject: [PATCH 25/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 96c68a2..4a3d104 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

From b93bbdc2a98b5b5bce166ba6febf35799d3387fe Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:05:06 +0800 Subject: [PATCH 26/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 4a3d104..3a9d0c3 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

From 0dcc7cc76eb6a660d292d7bf3d4b83a5b9a5457e Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:07:04 +0800 Subject: [PATCH 27/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 3a9d0c3..e963a34 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

+

[Homepage] | 🤖 [Chat with DeepSeek Coder] | 🤗 [Models Download] | [Discord] | [WeChat (微信)]

From 2bf0de3b921042e6451a257112495c4a5d416168 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:09:10 +0800 Subject: [PATCH 28/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index e963a34..b284cc7 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[Homepage] | 🤖 [Chat with DeepSeek Coder] | 🤗 [Models Download] | [Discord] | [WeChat (微信)]

+

[Homepage] | 🤖 [Chat with DeepSeek Coder] | 🤗 [Models Download] | [Discord] | [WeChat (微信)]

From b2e0f2916b3e53b830754c5a9611ed3d081ddce9 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:09:52 +0800 Subject: [PATCH 29/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index b284cc7..0b3b6c6 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@

DeepSeek Coder

-

[Homepage] | 🤖 [Chat with DeepSeek Coder] | 🤗 [Models Download] | [Discord] | [WeChat (微信)]

+

[ Homepage] | [🤖 Chat with DeepSeek Coder] | [🤗 Models Download] | [Discord] | [WeChat (微信)]

From 33c230697b1e6a1b218a286ca6ebeb3af6c629c2 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:24:07 +0800 Subject: [PATCH 30/35] Update README.md --- README.md | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index 0b3b6c6..3044b0f 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,7 @@ ### 1. Introduction of DeepSeek Coder -Deepseek Coder is composed of a series of code language models, each trained from scratch on 2T tokens, with a composition of 87% code and 13% natural language in both English and Chinese. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and an extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, Deepseek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks. +DeepSeek Coder is composed of a series of code language models, each trained from scratch on 2T tokens, with a composition of 87% code and 13% natural language in both English and Chinese. We provide various sizes of the code model, ranging from 1B to 33B versions. Each model is pre-trained on project-level code corpus by employing a window size of 16K and an extra fill-in-the-blank task, to support project-level code completion and infilling. For coding capabilities, DeepSeek Coder achieves state-of-the-art performance among open-source code models on multiple programming languages and various benchmarks.

result @@ -42,7 +42,7 @@ More evaluation details can be found in the [Detailed Evaluation](#5-detailed-ev #### Data Creation -- Step 1: Collecting code data from GitHub and apply the same filtering rules as [StarcoderData](https://github.com/bigcode-project/bigcode-dataset) to filter data. +- Step 1: Collecting code data from GitHub and apply the same filtering rules as [StarCoder Data](https://github.com/bigcode-project/bigcode-dataset) to filter data. - Step 2: Parsing the dependencies of files within the same repository to rearrange the file positions based on their dependencies. - Step 3: Concatenating dependent files to form a single example and employ repo-level minhash for deduplication. - Step 4: Further filtering out low-quality code, such as codes with syntax errors or poor readability. @@ -157,7 +157,7 @@ This code works by selecting a 'pivot' element from the array and partitioning t If you don't want to use the provided api `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. Replace the `['content']` with your instructions and the model's previous (if any) responses, then the model will generate the response to the currently given instruction. ``` -You are an AI programming assistant, utilizing the Deepseek Coder model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer. +You are an AI programming assistant, utilizing the DeepSeek Coder model, developed by DeepSeek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer. ### Instruction: ['content'] ### Response: @@ -255,7 +255,7 @@ print(tokenizer.decode(outputs[0])) ``` --- -In the following scenario, the Deepseek-Coder 6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in`main.py` file for model training and evaluation. +In the following scenario, the DeepSeek-Coder 6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in`main.py` file for model training and evaluation. ![Completion GIF](pictures/completion_demo.gif) From e1cc9ae6d90ca6775f8488c92a4f42b37a055fb9 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:27:10 +0800 Subject: [PATCH 31/35] Update README.md --- README.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index 3044b0f..cf02127 100644 --- a/README.md +++ b/README.md @@ -203,7 +203,7 @@ def load_data(): def evaluate_predictions(y_test, y_pred): return accuracy_score(y_test, y_pred) -#model.py +# model.py import torch import torch.nn as nn import torch.optim as optim @@ -242,7 +242,7 @@ class IrisClassifier(nn.Module): outputs = self(X_test) _, predicted = outputs.max(1) return predicted.numpy() -#main.py +# main.py from utils import load_data, evaluate_predictions from model import IrisClassifier as Classifier @@ -255,7 +255,7 @@ print(tokenizer.decode(outputs[0])) ``` --- -In the following scenario, the DeepSeek-Coder 6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in`main.py` file for model training and evaluation. +In the following scenario, the DeepSeek-Coder-6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in `main.py` file for model training and evaluation. ![Completion GIF](pictures/completion_demo.gif) From e593c64f7b3f49038d43f5f67fd01f994c2fc253 Mon Sep 17 00:00:00 2001 From: Chenggang Zhao Date: Mon, 6 Nov 2023 00:29:00 +0800 Subject: [PATCH 32/35] Update README.md --- README.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/README.md b/README.md index cf02127..e2ae554 100644 --- a/README.md +++ b/README.md @@ -203,6 +203,8 @@ def load_data(): def evaluate_predictions(y_test, y_pred): return accuracy_score(y_test, y_pred) + + # model.py import torch import torch.nn as nn @@ -242,6 +244,8 @@ class IrisClassifier(nn.Module): outputs = self(X_test) _, predicted = outputs.max(1) return predicted.numpy() + + # main.py from utils import load_data, evaluate_predictions from model import IrisClassifier as Classifier From 21aacd3e4b350641e2c2fe1dd561f5f67bb31efc Mon Sep 17 00:00:00 2001 From: Aleks B Date: Tue, 7 Nov 2023 17:11:21 +1030 Subject: [PATCH 33/35] Fix typos and grammatical errors in README.md --- README.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index e2ae554..ddf0057 100644 --- a/README.md +++ b/README.md @@ -24,7 +24,7 @@ DeepSeek Coder is composed of a series of code language models, each trained fro ### 2. Evaluation Results We evaluate DeepSeek Coder on various coding-related benchmarks. -Only `pass@1` results on HumanEval (Python and Multilingual), MBPP, DS-1000 are reported here: +Only `pass@1` results on HumanEval (Python and Multilingual), MBPP, and DS-1000 are reported here:

table @@ -33,7 +33,7 @@ Only `pass@1` results on HumanEval (Python and Multilingual), MBPP, DS-1000 are The result shows that DeepSeek-Coder-Base-33B significantly outperforms existing open-source code LLMs. Compared with CodeLlama-34B, it leads by 7.9%, 9.3%, 10.8% and 5.9% respectively on HumanEval Python, HumanEval Multilingual, MBPP and DS-1000. Surprisingly, our DeepSeek-Coder-Base-7B reaches the performance of CodeLlama-34B. -And the DeepSeek-Coder-Instruct-33B model after instruction tuning outperforms GPT35-turbo on HumanEval and achieves comparable result with GPT35-turbo on MBPP. +The DeepSeek-Coder-Instruct-33B model after instruction tuning outperforms GPT35-turbo on HumanEval and achieves comparable results with GPT35-turbo on MBPP. More evaluation details can be found in the [Detailed Evaluation](#5-detailed-evaluation-results). @@ -42,7 +42,7 @@ More evaluation details can be found in the [Detailed Evaluation](#5-detailed-ev #### Data Creation -- Step 1: Collecting code data from GitHub and apply the same filtering rules as [StarCoder Data](https://github.com/bigcode-project/bigcode-dataset) to filter data. +- Step 1: Collect code data from GitHub and apply the same filtering rules as [StarCoder Data](https://github.com/bigcode-project/bigcode-dataset) to filter data. - Step 2: Parsing the dependencies of files within the same repository to rearrange the file positions based on their dependencies. - Step 3: Concatenating dependent files to form a single example and employ repo-level minhash for deduplication. - Step 4: Further filtering out low-quality code, such as codes with syntax errors or poor readability. @@ -51,7 +51,7 @@ More evaluation details can be found in the [Detailed Evaluation](#5-detailed-ev #### Model Training -- Step 1: Initially pre-trained with a dataset consisting of 87% code, 10% code-related language (Github Markdown and StackExchange), and 3% non-code related Chinese language. Models are pre-trained using 1.8T tokens and a 4K window size in this step. +- Step 1: Initially pre-trained with a dataset consisting of 87% code, 10% code-related language (Github Markdown and StackExchange), and 3% non-code-related Chinese language. Models are pre-trained using 1.8T tokens and a 4K window size in this step. - Step 2: Further Pre-training using an extended 16K window size on an additional 200B tokens, resulting in foundational models (**DeepSeek-Coder-Base**). - Step 3: Instruction Fine-tuning on 2B tokens of instruction data, resulting in instruction-tuned models (**DeepSeek-Coder-Instruct**). @@ -63,7 +63,7 @@ Before proceeding, you'll need to install the necessary dependencies. You can do ``` pip install -r requirements.txt ``` -A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-33b-instruct), and you can run the demo locally using `app.py` in [demo](https://github.com/deepseek-ai/deepseek-coder/tree/main/demo) folder. (Thanks to all the HF team for their support) +A demo is also available on the [🤗 Hugging Face Space](https://huggingface.co/spaces/deepseek-ai/deepseek-coder-33b-instruct), and you can run the demo locally using `app.py` in the [demo](https://github.com/deepseek-ai/deepseek-coder/tree/main/demo) folder. (Thanks to all the HF team for their support) Here are some examples of how to use our model. @@ -155,7 +155,7 @@ print("Sorted array:", quick_sort(arr)) This code works by selecting a 'pivot' element from the array and partitioning the other elements into two sub-arrays, according to whether they are less than or greater than the pivot. The pivot element is then in its final position. The process is then repeated for the sub-arrays. ``` -If you don't want to use the provided api `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. Replace the `['content']` with your instructions and the model's previous (if any) responses, then the model will generate the response to the currently given instruction. +If you don't want to use the provided API `apply_chat_template` which loads the template from `tokenizer_config.json`, you can use the following template to chat with our model. Replace the `['content']` with your instructions and the model's previous (if any) responses, then the model will generate the response to the currently given instruction. ``` You are an AI programming assistant, utilizing the DeepSeek Coder model, developed by DeepSeek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer. ### Instruction: @@ -259,7 +259,7 @@ print(tokenizer.decode(outputs[0])) ``` --- -In the following scenario, the DeepSeek-Coder-6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in `main.py` file for model training and evaluation. +In the following scenario, the DeepSeek-Coder-6.7B model effectively calls a class **IrisClassifier** and its member function from the `model.py` file, and also utilizes functions from the `utils.py` file, to correctly complete the **main** function in the `main.py` file for model training and evaluation. ![Completion GIF](pictures/completion_demo.gif) From c5313cee6eadc1730b096ad597bbf0d0cca99905 Mon Sep 17 00:00:00 2001 From: Fuli Luo Date: Tue, 7 Nov 2023 17:00:29 +0800 Subject: [PATCH 34/35] Update README.md --- README.md | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index ddf0057..927b74a 100644 --- a/README.md +++ b/README.md @@ -278,12 +278,14 @@ The reproducible code for the following evaluation results can be found in the [ #### 4) Program-Aid Math Reasoning Benchmark ![Math](pictures/Math.png) +### 6. Resources +(awesome-deepseek-coder)[https://github.com/deepseek-ai/awesome-deepseek-coder] is a curated list of open-source projects related to DeepSeek Coder. -### 6. License +### 7. License This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the Model License. DeepSeek Coder supports commercial use. See the [LICENSE-CODE](LICENSE-CODE) and [LICENSE-MODEL](LICENSE-MODEL) for more details. -### 6. Contact +### 8. Contact If you have any questions, please raise an issue or contact us at [agi_code@deepseek.com](mailto:agi_code@deepseek.com). From d31ad8e67fdf0ba1260e87ed2f74b2e307e60b83 Mon Sep 17 00:00:00 2001 From: Fuli Luo Date: Tue, 7 Nov 2023 17:01:08 +0800 Subject: [PATCH 35/35] Update README.md --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 927b74a..4145452 100644 --- a/README.md +++ b/README.md @@ -279,7 +279,7 @@ The reproducible code for the following evaluation results can be found in the [ ![Math](pictures/Math.png) ### 6. Resources -(awesome-deepseek-coder)[https://github.com/deepseek-ai/awesome-deepseek-coder] is a curated list of open-source projects related to DeepSeek Coder. +[awesome-deepseek-coder](https://github.com/deepseek-ai/awesome-deepseek-coder) is a curated list of open-source projects related to DeepSeek Coder. ### 7. License This code repository is licensed under the MIT License. The use of DeepSeek Coder models is subject to the Model License. DeepSeek Coder supports commercial use.