license: openrail
datasets:
- bigcode/the-stack
language:
- code
programming_language:
- Java
- JavaScript
- Python
pipeline_tag: text-generation
widget:
- text: 'def print_hello_world():'
example_title: Hello world
group: Python
model-index:
- name: SantaCoder
results:
- task:
type: text-generation
dataset:
type: openai_humaneval
name: HumanEval (Python)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
verified: false
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL HumanEval (Java)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL HumanEval (JavaScript)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: openai_humaneval
name: MBPP (Python)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
verified: false
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL MBPP (Java)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL MBPP (JavaScript)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: loubnabnl/humaneval_infilling
name: HumanEval FIM (Python)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
verified: false
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL HumanEval FIM (Java)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: nuprl/MultiPL-E
name: MultiPL HumanEval FIM (JavaScript)
metrics:
- name: pass@1
type: pass@1
value: 0
verified: false
- name: pass@10
type: pass@10
value: 0
verified: false
- name: pass@100
type: pass@100
value: 0
- task:
type: text-generation
dataset:
type: code_x_glue_ct_code_to_text
name: CodeXGLUE code-to-text (Python)
metrics:
- name: BLEU
type: bleu
value: 0
verified: false
SantaCoder
Table of Contents
Model Summary
The SantaCoder models are a series of 1B parameter models trained on the Python, Java, and JavaScript subset of The Stack (v1.1) (which excluded opt-out requests). The main model uses multi-query attention, was trained using near-deduplication and comment-to-code ratio as filtering criteria and using the Fill-in-the-Middle objective. In addition there are several models that were trained on datasets with different filter parameters and with architecture and objective variations.
- Repository: bigcode/Megatron-LM
- Project Website: bigcode-project.org
- Paper: 🎅SantaCoder: Don't reach for the stars!🌟
- Point of Contact: contact@bigcode-project.org
- Languages: Python, Java, and JavaScript
Model | Architecture | Objective | Filtering |
---|---|---|---|
mha |
MHA | AR + FIM | Base |
no-fim |
MQA | AR | Base |
fim |
MQA | AR + FIM | Base |
stars |
MQA | AR + FIM | GitHub stars |
fertility |
MQA | AR + FIM | Tokenizer fertility |
comments |
MQA | AR + FIM | Comment-to-code ratio |
dedup-alt |
MQA | AR + FIM | Stronger near-deduplication |
dedup-alt-comments |
MQA | AR + FIM | Stronger near-deduplication and comment-to-code ratio |
The dedup-alt-comments
model is the best performing model and was trained twice as long as the others. This checkpoint is the default model and available on the main
branch. All other checkpoints are on separate branches with according names.
Use
Intended use
The model was trained on GitHub code. As such it is not an instruction model and commands like "Write a function that computes the square root." do not work well.
You should phrase commands like they occur in source code such as comments (e.g. # the following function computes the sqrt
) or write a function signature and docstring and let the model complete the function body.
Feel free to share your generations in the Community tab!
How to use
Generation
# pip install -q transformers
from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "bigcode/santacoder"
device = "cuda" # for GPU usage or "cpu" for CPU usage
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForCausalLM.from_pretrained(checkpoint, trust_remote_code=True).to()
inputs = tokenizer.encode("def print_hello_world():", return_tensors="pt").to(device)
outputs = model.generate(inputs)
print(tokenizer.decode(outputs[0]))
Fill-in-the-middle
Fill-in-the-mid uses special tokens to identify the prefix/middle/suffic part of the input and output:
input_text = "<fim-prefix>def print_hello_world():\n <fim-suffix>\n print('Hello world!')<fim-middle>"
inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)
outputs = model.generate(inputs)
print(tokenizer.decode(outputs[0]))
Load other checkpoints
We upload the checkpoint of each experiment to a seperate branch as well as the intermediate checkpoints as commits on the branches. You can load them with the revision
flag:
model = AutoModelForCausalLM.from_pretrained(
"bigcode/santacoder",
revision="no-fim", # name of branch or commit hash
trust_remote_code=True
)
Attribution
The pretraining dataset of the model was filtered for permissive licenses only. Nevertheless, the model can generate source code verbatim from the dataset which requires attribution. We provide a search index that let's you search through the pretraining data to identify where generated code came from and apply the proper attribution to your code.
Limitations
The model has been trained on source code in Python, Java, and JavaScript. The predominant language in source is English although other languages are also present. As such the model is capable to generate code snippets provided some context but the generated code is not guaranteed to work as intended. It can be inefficient, contain bugs or exploits.
Training
Model
- Architecture: GPT-2 model with multi-query attention and Fill-in-the-Middle objective
- Pretraining steps: 600K
- Pretraining tokens: 236 billion
- Precision: float16
Hardware
- GPUs: 96 Tesla V100
- Training time: 6.2 days
- Total FLOPS: 2.1 x 10e21
Software
- Orchestration: Megatron-LM
- Neural networks: PyTorch
- FP16 if applicable: apex
License
The model is licenses under the CodeML Open RAIL-M v0.1 license. You can find the full license here.
Citation
TODO