--- pipeline_tag: text-generation inference: true widget: - text: 'def print_hello_world():' example_title: Hello world group: Python datasets: - bigcode/the-stack-v2-train license: bigcode-openrail-m library_name: transformers tags: - code model-index: - name: starcoder2-7b results: - task: type: text-generation dataset: name: CruxEval-I type: cruxeval-i metrics: - type: pass@1 value: 34.6 - task: type: text-generation dataset: name: DS-1000 type: ds-1000 metrics: - type: pass@1 value: 27.8 - task: type: text-generation dataset: name: GSM8K (PAL) type: gsm8k-pal metrics: - type: accuracy value: 40.4 - task: type: text-generation dataset: name: HumanEval+ type: humanevalplus metrics: - type: pass@1 value: 29.9 - task: type: text-generation dataset: name: HumanEval type: humaneval metrics: - type: pass@1 value: 35.4 - task: type: text-generation dataset: name: RepoBench-v1.1 type: repobench-v1.1 metrics: - type: edit-smiliarity value: 72.07 --- # StarCoder2
SC2
## Table of Contents 1. [Model Summary](##model-summary) 2. [Use](##use) 3. [Limitations](##limitations) 4. [Training](##training) 5. [License](##license) 6. [Citation](##citation) ## Model Summary StarCoder2-7B model is a 7B parameter model trained on 17 programming languages from [The Stack v2](https://huggingface.co/datasets/bigcode/the-stack-v2-train), with opt-out requests excluded. The model uses [Grouped Query Attention](https://arxiv.org/abs/2305.13245), [a context window of 16,384 tokens](https://arxiv.org/abs/2205.14135) with [a sliding window attention of 4,096 tokens](https://arxiv.org/abs/2004.05150v2), and was trained using the [Fill-in-the-Middle objective](https://arxiv.org/abs/2207.14255) on 3.5+ trillion tokens. - **Project Website:** [bigcode-project.org](https://www.bigcode-project.org) - **Paper:** [Link](https://huggingface.co/datasets/bigcode/the-stack-v2/) - **Point of Contact:** [contact@bigcode-project.org](mailto:contact@bigcode-project.org) - **Languages:** 17 Programming languages ## Use ### Intended use The model was trained on GitHub code as well as additional selected data sources such as Arxiv and Wikipedia. As such it is _not_ an instruction model and commands like "Write a function that computes the square root." do not work well. ### Generation Here are some examples to get started with the model. You can find a script for fine-tuning in StarCoder2's [GitHub repository](https://github.com/bigcode-project/starcoder2). First, make sure to install `transformers` from source: ```bash pip install git+https://github.com/huggingface/transformers.git ``` #### Running the model on CPU/GPU/multi GPU * _Using full precision_ ```python # pip install git+https://github.com/huggingface/transformers.git # TODO: merge PR to main from transformers import AutoModelForCausalLM, AutoTokenizer checkpoint = "bigcode/starcoder2-7b" device = "cuda" # for GPU usage or "cpu" for CPU usage tokenizer = AutoTokenizer.from_pretrained(checkpoint) # for multiple GPUs install accelerate and do `model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map="auto")` model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device) inputs = tokenizer.encode("def print_hello_world():", return_tensors="pt").to(device) outputs = model.generate(inputs) print(tokenizer.decode(outputs[0])) ``` ```bash >>> print(f"Memory footprint: {model.get_memory_footprint() / 1e6:.2f} MB") Memory footprint: 29232.57 MB ``` * _Using `torch.bfloat16`_ ```python # pip install accelerate import torch from transformers import AutoTokenizer, AutoModelForCausalLM checkpoint = "bigcode/starcoder2-7b" tokenizer = AutoTokenizer.from_pretrained(checkpoint) # for fp16 use `torch_dtype=torch.float16` instead model = AutoModelForCausalLM.from_pretrained(checkpoint, device_map="auto", torch_dtype=torch.bfloat16) inputs = tokenizer.encode("def print_hello_world():", return_tensors="pt").to("cuda") outputs = model.generate(inputs) print(tokenizer.decode(outputs[0])) ``` ```bash >>> print(f"Memory footprint: {model.get_memory_footprint() / 1e6:.2f} MB") Memory footprint: 14616.29 MB ``` #### Quantized Versions through `bitsandbytes` * _Using 8-bit precision (int8)_ ```python # pip install bitsandbytes accelerate from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig # to use 4bit use `load_in_4bit=True` instead quantization_config = BitsAndBytesConfig(load_in_8bit=True) checkpoint = "bigcode/starcoder2-7b" tokenizer = AutoTokenizer.from_pretrained(checkpoint) model = AutoModelForCausalLM.from_pretrained(checkpoint, quantization_config=quantization_config) inputs = tokenizer.encode("def print_hello_world():", return_tensors="pt").to("cuda") outputs = model.generate(inputs) print(tokenizer.decode(outputs[0])) ``` ```bash >>> print(f"Memory footprint: {model.get_memory_footprint() / 1e6:.2f} MB") # load_in_8bit Memory footprint: 7670.52 MB # load_in_4bit >>> print(f"Memory footprint: {model.get_memory_footprint() / 1e6:.2f} MB") Memory footprint: 4197.64 MB ``` ### Attribution & Other Requirements The pretraining dataset of the model was filtered for permissive licenses and code with no license only. Nevertheless, the model can generate source code verbatim from the dataset. The code's license might require attribution and/or other specific requirements that must be respected. We provide a [search index](TODO) that lets you search through the pretraining data to identify where the generated code came from and apply the proper attribution to your code. # Limitations The model has been trained on source code from 600+ programming languages. The predominant language in source is English although other languages are also present. As such the model is capable of generating code snippets provided some context but the generated code is not guaranteed to work as intended. It can be inefficient and contain bugs or exploits. See [the paper](TODO) for an in-depth discussion of the model limitations. # Training ## Model - **Architecture:** Transformer decoder with grouped-query and sliding window attention and Fill-in-the-Middle objective - **Pretraining steps:** 1 million - **Pretraining tokens:** 3.5+ trillion - **Precision:** bfloat16 ## Hardware - **GPUs:** 432 H100 ## Software - **Framework:** [nanotron](https://github.com/huggingface/nanotron/) - **Neural networks:** [PyTorch](https://github.com/pytorch/pytorch) # License The model is licensed under the BigCode OpenRAIL-M v1 license agreement. You can find the full agreement [here](https://huggingface.co/spaces/bigcode/bigcode-model-license-agreement). # Citation _Coming soon_