madhavatreplit's picture
Update README.md (#8)
7ae75a4
|
raw
history blame
4.56 kB
metadata
license: apache-2.0
datasets:
  - bigcode/the-stack-dedup
  - togethercomputer/RedPajama-Data-1T
tags:
  - code
  - Composer
  - MosaicML
  - llm-foundry
  - StreamingDatasets
language:
  - code

Replit Code V-1.5 3B

Developed by: Replit, Inc.

Model Description

Replit Code v1.5 is a 3.3B parameter Causal Language Model focused on Code Completion.

The model is trained in bfloat16 on 1T tokens of code (~200B tokens over 5 epochs, including linear cooldown) for 30 programming languages from a subset of permissively licensed code from Bigcode's Stack Dedup dataset, a filtered natural language sample from Markdown and reStructuredText subsets from the same Stack Dedup dataset, and a dev-oriented sample from RedPajama's StackExchange dataset sourced from the Stack Exchange Data Dump by Stack Exchange Inc.

The 30 programming languages are:

Java, JavaScript, C, PHP, Python, C++, C#, TypeScript, Go, CSS, HTML, Rust, Ruby, Swift, Scala, Shell, Lua, Perl, Haskell, JSX, Julia, Common Lisp, OCaml, Solidity, Scheme, R, Zig, SQL, Racket, D

The context size of the model is 4096 tokens. We use the GPTNeoX tokenizer with a custom trained and optimized vocabulary of 32768 tokens. This custom vocabulary led to single-digit % points on compression while maintaining or improving coverage on our training corpus.

The model has been trained on the MosaicML platform on 128 H100-80GB GPUs using their LLM Foundry and Composer training library built on top of PyTorch.

Dependencies

You will need to install the latest versions of the following dependencies:

einops
torch
transformers

How to Use

Generation

You can generate code using the transformers library as follows:

from transformers import AutoModelForCausalLM, AutoTokenizer

tokenizer = AutoTokenizer.from_pretrained('replit/replit-code-v1_5-3b', trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained('replit/replit-code-v1_5-3b', trust_remote_code=True)

x = tokenizer.encode('def fibonacci(n): ', return_tensors='pt')
y = model.generate(x, max_length=100, do_sample=True, top_p=0.95, top_k=4, temperature=0.2, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id)

# decoding
generated_code = tokenizer.decode(y[0], skip_special_tokens=True, clean_up_tokenization_spaces=False)
print(generated_code)

Experiment with different decoding methods and parameters to get the best results for your use case.

Using Triton Implementation of Flash Attention

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig

config = AutoConfig.from_pretrained(
    "replit/replit-code-v1_5-3b",
    trust_remote_code=True
)
config.attn_config['attn_impl'] = 'triton'

# load model
tokenizer = AutoTokenizer.from_pretrained('replit/replit-code-v1_5-3b', trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained('replit/replit-code-v1_5-3b', config=config, trust_remote_code=True)
model.to(device='cuda:0', dtype=torch.bfloat16)

# forward pass
x = tokenizer.encode('def fibonacci(n): ', return_tensors='pt').to(device='cuda:0')
x = x.to(device='cuda:0')
y = model.generate(x, max_length=100, do_sample=True, top_p=0.95, top_k=4, temperature=0.2, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id)


# decoding
generated_code = tokenizer.decode(y[0], skip_special_tokens=True, clean_up_tokenization_spaces=False)
print(generated_code)

Experiment with different decoding methods and parameters to get the best results for your use case. We recommend experimenting with temperature and reptition_penaltyfor optimal performance on your use case!

Intended Use

Replit intends this model be used by anyone as a foundational model for application-specific fine-tuning without strict limitations on commercial use.

The model is trained specifically for code completion tasks.

Limitations

The pre-training dataset may have contained offensive or inappropriate content even after applying data cleansing and toxicity and profanity filters, and such content may be reflected in model generated text. We recommend that users exercise reasonable caution when using in production systems. Do not use for any applications that may cause harm or distress to individuals or groups.