I am continuously enhancing the structure of these model descriptions, and they now provide even more comprehensive information to help you find the best models for your specific needs.
falcon-40b - GGUF
- Model creator: tiiuae
- Original model: falcon-40b
Note: Important Update for Falcon Models in llama.cpp Versions After October 18, 2023
As noted on the [Llama.cpp](ggerganov/llama.cpp: Port of Facebook's LLaMA model in C/C++ (github.com) GitHub repository, all new releases of Llama.cpp will require a re-quantization due to the implementation of the new BPE tokenizer, which impacts both the original Falcon models and their derived variants.
Here's what you need to know:
Original Falcon Models: I am diligently working to provide updated quantized versions of the four original Falcon models to ensure their compatibility with the new llama.cpp versions. Please keep an eye on my Hugging Face Model pages for updates on the availability of these models. Promptly downloading them is essential to maintain compatibility with the latest llama.cpp releases.
Derived Falcon Models: Right now, the derived Falcon-Models cannot be re-converted without adjustments from the original model creators. So far, these models cannot be used in recent llama.cpp versions at all. ** Good news!** It's in the pipeline that the capability for quantizing even the older derived Falcon models will be incorporated soon. However, the exact timeline is beyond my control.
Stay Informed: Application software using llama.cpp libraries will follow soon. Keep an eye on the release schedules of your favorite software applications that rely on llama.cpp. They will likely provide instructions on how to integrate the new models.
Monitor Upload Times: Please keep a close watch on the upload times of the available files on my Hugging Face Model pages. This will help you identify which files have already been updated and are ready for download, ensuring you have the most current Falcon models at your disposal.
Download Promptly: Once the updated Falcon models are available on my Hugging Face page, be sure to download them promptly to ensure compatibility with the latest [llama.cpp](ggerganov/llama.cpp: Port of Facebook's LLaMA model in C/C++ (github.com) versions.
Please understand that this change specifically affects Falcon and Starcoder models, other models remain unaffected. Consequently, software providers may not emphasize this change as prominently.
As a solo operator of this page, I'm doing my best to expedite the process, but please bear with me as this may take some time.
These are gguf quantized models of the riginal Falcon 40B Model by tiiuae. Falcon is a foundational large language model coming in different sizes: 7b, 40b and 180b. Sadly, as the Falcon 180b Models are note really free models, I do not provide quantized versions here.
About GGUF format
gguf
is the current file format used by the ggml
library.
A growing list of Software is using it and can therefore use this model.
The core project making use of the ggml library is the llama.cpp project by Georgi Gerganov
Quantization variants
There is a bunch of quantized files available. How to choose the best for you:
legacy quants
Q4_0, Q4_1, Q5_0, Q5_1 and Q8 are legacy
quantization types.
Nevertheless, they are fully supported, as there are several circumstances that cause certain model not to be compatible with the modern K-quants.
Falcon 7B models cannot be quantized to K-quants.
K-quants
K-quants are based on the idea that the quantization of certain parts affects the quality in different ways. If you quantize certain parts more and others less, you get a more powerful model with the same file size, or a smaller file size and lower memory load with comparable performance. So, if possible, use K-quants. With a Q6_K you should find it really hard to find a quality difference to the original model - ask your model two times the same question and you may encounter bigger quality differences.
Original Model Card:
π 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 π.
π€ 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.
- β οΈ 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.
πΈ 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']}")
π₯ 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).
Data source | Fraction | Tokens | Sources |
---|---|---|---|
RefinedWeb-English | 75% | 750B | massive web crawl |
RefinedWeb-Europe | 7% | 70B | European massive web crawl |
Books | 6% | 60B | |
Conversations | 5% | 50B | Reddit, StackOverflow, HackerNews |
Code | 5% | 50B | |
Technical | 2% | 20B | arXiv, PubMed, USPTO, etc. |
RefinedWeb-Europe is made of the following languages:
Language | Fraction of multilingual data | Tokens |
---|---|---|
German | 26% | 18B |
Spanish | 24% | 17B |
French | 23% | 16B |
Italian | 7% | 5B |
Portuguese | 4% | 3B |
Polish | 4% | 3B |
Dutch | 4% | 3B |
Romanian | 3% | 2B |
Czech | 3% | 2B |
Swedish | 2% | 1B |
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
Hyperparameter | Value | Comment |
---|---|---|
Precision | bfloat16 |
|
Optimizer | AdamW | |
Learning rate | 1.85e-4 | 4B tokens warm-up, cosine decay to 1.85e-5 |
Weight decay | 1e-1 | |
Z-loss | 1e-4 | |
Batch size | 1152 | 100B tokens ramp-up |
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.
Hyperparameter | Value | Comment |
---|---|---|
Layers | 60 | |
d_model |
8192 | |
head_dim |
64 | Reduced to optimise for FlashAttention |
Vocabulary | 65024 | |
Sequence length | 2048 |
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 π. 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 π 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
End of original Model File
Please consider to support my work
Coming Soon: I'm in the process of launching a sponsorship/crowdfunding campaign for my work. I'm evaluating Kickstarter, Patreon, or the new GitHub Sponsors platform, and I am hoping for some support and contribution to the continued availability of these kind of models. Your support will enable me to provide even more valuable resources and maintain the models you rely on. Your patience and ongoing support are greatly appreciated as I work to make this page an even more valuable resource for the community.
- Downloads last month
- 863