ct2fast-starchat-beta

# Fast-Inference with Ctranslate2

Speedup inference while reducing memory by 2x-4x using int8 inference in C++ on CPU or GPU.

quantized version of HuggingFaceH4/starchat-beta

pip install hf-hub-ctranslate2>=2.0.8 ctranslate2>=3.14.0

Converted on 2023-06-12 using

ct2-transformers-converter --model HuggingFaceH4/starchat-beta --output_dir /home/michael/tmp-ct2fast-starchat-beta --force --copy_files merges.txt all_results.json training_args.bin tokenizer.json README.md dialogue_template.json tokenizer_config.json eval_results.json vocab.json train_results.json generation_config.json trainer_state.json special_tokens_map.json added_tokens.json requirements.txt .gitattributes --quantization int8_float16 --trust_remote_code

Checkpoint compatible to ctranslate2>=3.15.0 and hf-hub-ctranslate2>=2.0.8

• compute_type=int8_float16 for device="cuda"
• compute_type=int8 for device="cpu"

from hf_hub_ctranslate2 import TranslatorCT2fromHfHub, GeneratorCT2fromHfHub
from transformers import AutoTokenizer

model_name = "michaelfeil/ct2fast-starchat-beta"
# use either TranslatorCT2fromHfHub or GeneratorCT2fromHfHub here, depending on model.
model = GeneratorCT2fromHfHub(
        # load in int8 on CUDA
        model_name_or_path=model_name,
        device="cuda",
        compute_type="int8_float16",
        # tokenizer=AutoTokenizer.from_pretrained("HuggingFaceH4/starchat-beta")
)
outputs = model.generate(
    text=["def fibonnaci(", "User: How are you doing? Bot:"],
    max_length=64,
    include_prompt_in_result=False
)
print(outputs)

Licence and other remarks:

This is just a quantized version. Licence conditions are intended to be idential to original huggingface repo.

Original description

<img src="https://huggingface.co/HuggingFaceH4/starchat-beta/resolve/main/model_logo.png" alt="StarChat Beta Logo" width="800" style="margin-left:'auto' margin-right:'auto' display:'block'"/>

Model Card for StarChat-β

StarChat is a series of language models that are trained to act as helpful coding assistants. StarChat-β is the second model in the series, and is a fine-tuned version of StarCoderPlus that was trained on an "uncensored" variant of the openassistant-guanaco dataset. We found that removing the in-built alignment of the OpenAssistant dataset boosted performance on the Open LLM Leaderboard and made the model more helpful at coding tasks. However, this means that model is likely to generate problematic text when prompted to do so and should only be used for educational and research purposes.

Model Details

Model Description

<!-- Provide a longer summary of what this model is. -->

• Model type: A 16B parameter GPT-like model fine-tuned on an "uncensored" variant of the openassistant-guanaco dataset.
• Language(s) (NLP): Primarily English and 80+ programming languages.
• License: BigCode Open RAIL-M v1
• Finetuned from model: bigcode/starcoderplus

Model Sources

<!-- Provide the basic links for the model. -->

• Repository: https://github.com/bigcode-project/starcoder
• Demo: https://huggingface.co/spaces/HuggingFaceH4/starchat-playground

Intended uses & limitations

The model was fine-tuned on a variant of the OpenAssistant/oasst1 dataset, which contains a diverse range of dialogues in over 35 languages. As a result, the model can be used for chat and you can check out our demo to test its coding capabilities.

Here's how you can run the model using the pipeline() function from 🤗 Transformers:

import torch
from transformers import pipeline

pipe = pipeline("text-generation", model="HuggingFaceH4/starchat-beta", torch_dtype=torch.bfloat16, device_map="auto")

# We use a variant of ChatML to format each message
prompt_template = "<|system|>\n<|end|>\n<|user|>\n{query}<|end|>\n<|assistant|>"
prompt = prompt_template.format(query="How do I sort a list in Python?")
# We use a special <|end|> token with ID 49155 to denote ends of a turn
outputs = pipe(prompt, max_new_tokens=256, do_sample=True, temperature=0.2, top_k=50, top_p=0.95, eos_token_id=49155)
# You can sort a list in Python by using the sort() method. Here's an example:\n\n```\nnumbers = [3, 1, 4, 1, 5, 9, 2, 6, 5, 3, 5]\nnumbers.sort()\nprint(numbers)\n```\n\nThis will sort the list in place and print the sorted list.

Bias, Risks, and Limitations

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

StarChat-β has not been aligned to human preferences with techniques like RLHF or deployed with in-the-loop filtering of responses like ChatGPT, so the model can produce problematic outputs (especially when prompted to do so). Models trained primarily on code data will also have a more skewed demographic bias commensurate with the demographics of the GitHub community, for more on this see the StarCoder dataset which is derived from The Stack.

Since the base model was pretrained on a large corpus of code, it may produce code snippets that are syntactically valid but semantically incorrect. For example, it may produce code that does not compile or that produces incorrect results.
It may also produce code that is vulnerable to security exploits.
We have observed the model also has a tendency to produce false URLs which should be carefully inspected before clicking.

StarChat-β was fine-tuned from the base model StarCoderPlus, please refer to its model card's Limitations Section for relevant information. In particular, the model was evaluated on some categories of gender biases, propensity for toxicity, and risk of suggesting code completions with known security flaws; these evaluations are reported in its technical report.

Training and evaluation data

StarChat-β is trained on an "uncensored" variant of the openassistant-guanaco dataset. We applied the same recipe used to filter the ShareGPT datasets behind the WizardLM.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• total_eval_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 6

Training results

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3

Citation

Although there isn't a blog post or paper associated with StarChat-β, you can find details on the earlier version in the blog post below:

BibTeX:

@article{Tunstall2023starchat-alpha,
  author = {Tunstall, Lewis and Lambert, Nathan and Rajani, Nazneen and Beeching, Edward and Le Scao, Teven and von Werra, Leandro and Han, Sheon and Schmid, Philipp and Rush, Alexander},
  title = {Creating a Coding Assistant with StarCoder},
  journal = {Hugging Face Blog},
  year = {2023},
  note = {https://huggingface.co/blog/starchat},
}