michaelfeil / ct2fast-falcon-40b

# Fast-Inference with Ctranslate2

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

quantized version of tiiuae/falcon-40b

pip install hf-hub-ctranslate2>=2.10.0 ctranslate2>=3.16.0

# from transformers import AutoTokenizer
model_name = "michaelfeil/ct2fast-falcon-40b"

from hf_hub_ctranslate2 import GeneratorCT2fromHfHub
model = GeneratorCT2fromHfHub(
        # load in int8 on CUDA
        model_name_or_path=model_name,
        device="cuda",
        compute_type="int8_float16",
        # tokenizer=AutoTokenizer.from_pretrained("{ORG}/{NAME}")
)
outputs = model.generate(
    text=["def fibonnaci(", "User: How are you doing? Bot:"],
    max_length=64,
    include_prompt_in_result=False
)
print(outputs)

Checkpoint compatible to ctranslate2>=3.16.0 and hf-hub-ctranslate2>=2.10.0

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

Converted on 2023-06-16 using

ct2-transformers-converter --model tiiuae/falcon-40b --output_dir ~/tmp-ct2fast-falcon-40b --force --copy_files tokenizer.json README.md tokenizer_config.json generation_config.json special_tokens_map.json .gitattributes --quantization int8_float16 --trust_remote_code

Licence and other remarks:

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

Original description

🚀 Falcon-40B

Falcon-40B is a 40B parameters causal decoder-only model built by TII and trained on 1,000B tokens of RefinedWeb enhanced with curated corpora. It is made available under the Apache 2.0 license.

Paper coming soon 😊.

Call for Proposals : Falcon 40B - World's Top Ranked AI Model Empowers Exceptional Use Cases with Training Compute Power in Call for Proposals

We get it. AI is everywhere! Is it taking over?

Before we debate the scant likelihood of a cyborg assassin from the future terminating humanity, let’s get to know the newbie that has soared to top-spot on the leaderboard – Falcon 40B.

Falcon 40B is the UAE’s and the Middle East’s first home-grown, open-source large language model (LLM) with 40 billion parameters trained on one trillion tokens. The brainchild of the Technology Innovation Institute (TII), Falcon 40B has generated a tremendous amount of global interest and intrigue, but what really sweetens the deal is its transparent, open-source feature.

TII is now calling for proposals from users worldwide to submit their most creative ideas for Falcon 40B’s deployment – allowing them to share their knowledge, enhance the software, and potentially transform their ideas into reality! Take that, ChatGPT! Worth checking out? Give it a go and see for yourself!

Submit your proposal today! https://falconllm.tii.ae/call-for-proposal.php

🤗 To get started with Falcon (inference, finetuning, quantization, etc.), we recommend reading this great blogpost fron HF !

Why use Falcon-40B?

• It is the best open-source model currently available. Falcon-40B outperforms LLaMA , StableLM , RedPajama , MPT , etc. See the OpenLLM Leaderboard .
• It features an architecture optimized for inference , with FlashAttention ( Dao et al., 2022 ) and multiquery ( Shazeer et al., 2019 ).
• It is made available under a permissive Apache 2.0 license allowing for commercial use , without any royalties or restrictions.
• ⚠️ This is a raw, pretrained model, which should be further finetuned for most usecases. If you are looking for a version better suited to taking generic instructions in a chat format, we recommend taking a look at Falcon-40B-Instruct .

💸 Looking for a smaller, less expensive model? Falcon-7B is Falcon-40B's little brother!

from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model = "tiiuae/falcon-40b"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    device_map="auto",
)
sequences = pipeline(
   "Girafatron is obsessed with giraffes, the most glorious animal on the face of this Earth. Giraftron believes all other animals are irrelevant when compared to the glorious majesty of the giraffe.\nDaniel: Hello, Girafatron!\nGirafatron:",
    max_length=200,
    do_sample=True,
    top_k=10,
    num_return_sequences=1,
    eos_token_id=tokenizer.eos_token_id,
)
for seq in sequences:
    print(f"Result: {seq['generated_text']}")

💥 Falcon LLMs require PyTorch 2.0 for use with transformers !

For fast inference with Falcon, check-out Text Generation Inference ! Read more in this blogpost .

You will need at least 85-100GB of memory to swiftly run inference with Falcon-40B.

Model Card for Falcon-40B

Model Details

Model Description

• Developed by: https://www.tii.ae ;
• Model type: Causal decoder-only;
• Language(s) (NLP): English, German, Spanish, French (and limited capabilities in Italian, Portuguese, Polish, Dutch, Romanian, Czech, Swedish);
• License: Apache 2.0 license.

Model Source

• Paper: coming soon .

Uses

Direct Use

Research on large language models; as a foundation for further specialization and finetuning for specific usecases (e.g., summarization, text generation, chatbot, etc.)

Out-of-Scope Use

Production use without adequate assessment of risks and mitigation; any use cases which may be considered irresponsible or harmful.

Bias, Risks, and Limitations

Falcon-40B is trained mostly on English, German, Spanish, French, with limited capabilities also in in Italian, Portuguese, Polish, Dutch, Romanian, Czech, Swedish. It will not generalize appropriately to other languages. Furthermore, as it is trained on a large-scale corpora representative of the web, it will carry the stereotypes and biases commonly encountered online.

Recommendations

We recommend users of Falcon-40B to consider finetuning it for the specific set of tasks of interest, and for guardrails and appropriate precautions to be taken for any production use.

How to Get Started with the Model

from transformers import AutoTokenizer, AutoModelForCausalLM
import transformers
import torch

model = "tiiuae/falcon-40b"

tokenizer = AutoTokenizer.from_pretrained(model)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    torch_dtype=torch.bfloat16,
    trust_remote_code=True,
    device_map="auto",
)
sequences = pipeline(
   "Girafatron is obsessed with giraffes, the most glorious animal on the face of this Earth. Giraftron believes all other animals are irrelevant when compared to the glorious majesty of the giraffe.\nDaniel: Hello, Girafatron!\nGirafatron:",
    max_length=200,
    do_sample=True,
    top_k=10,
    num_return_sequences=1,
    eos_token_id=tokenizer.eos_token_id,
)
for seq in sequences:
    print(f"Result: {seq['generated_text']}")

Training Details

Training Data

Falcon-40B was trained on 1,000B tokens of RefinedWeb , a high-quality filtered and deduplicated web dataset which we enhanced with curated corpora. Significant components from our curated copora were inspired by The Pile ( Gao et al., 2020 ).

RefinedWeb-Europe is made of the following languages:

The data was tokenized with the Falcon- 7B / 40B tokenizer.

Training Procedure

Falcon-40B was trained on 384 A100 40GB GPUs, using a 3D parallelism strategy (TP=8, PP=4, DP=12) combined with ZeRO.

Training Hyperparameters

Speeds, Sizes, Times

Training started in December 2022 and took two months.

Evaluation

Paper coming soon.

See the OpenLLM Leaderboard for early results.

Technical Specifications

Model Architecture and Objective

Falcon-40B is a causal decoder-only model trained on a causal language modeling task (i.e., predict the next token).

The architecture is broadly adapted from the GPT-3 paper ( Brown et al., 2020 ), with the following differences:

• Positionnal embeddings: rotary ( Su et al., 2021 );
• Attention: multiquery ( Shazeer et al., 2019 ) and FlashAttention ( Dao et al., 2022 );
• Decoder-block: parallel attention/MLP with a two layer norms.

For multiquery, we are using an internal variant which uses independent key and values per tensor parallel degree.

Compute Infrastructure

Hardware

Falcon-40B was trained on AWS SageMaker, on 384 A100 40GB GPUs in P4d instances.

Software

Falcon-40B was trained a custom distributed training codebase, Gigatron. It uses a 3D parallelism approach combined with ZeRO and high-performance Triton kernels (FlashAttention, etc.)

Citation

Paper coming soon 😊. In the meanwhile, you can use the following information to cite:

@article{falcon40b,
  title={{Falcon-40B}: an open large language model with state-of-the-art performance},
  author={Almazrouei, Ebtesam and Alobeidli, Hamza and Alshamsi, Abdulaziz and Cappelli, Alessandro and Cojocaru, Ruxandra and Debbah, Merouane and Goffinet, Etienne and Heslow, Daniel and Launay, Julien and Malartic, Quentin and Noune, Badreddine and Pannier, Baptiste and Penedo, Guilherme},
  year={2023}
}

To learn more about the pretraining dataset, see the 📓 RefinedWeb paper .

@article{refinedweb,
  title={The {R}efined{W}eb dataset for {F}alcon {LLM}: outperforming curated corpora with web data, and web data only},
  author={Guilherme Penedo and Quentin Malartic and Daniel Hesslow and Ruxandra Cojocaru and Alessandro Cappelli and Hamza Alobeidli and Baptiste Pannier and Ebtesam Almazrouei and Julien Launay},
  journal={arXiv preprint arXiv:2306.01116},
  eprint={2306.01116},
  eprinttype = {arXiv},
  url={https://arxiv.org/abs/2306.01116},
  year={2023}
}

License

Falcon-40B is made available under the Apache 2.0 license.

Contact

[email protected]