falcon-mamba

0. TL;DR
1. Model Details
2. Usage
3. Training Details
4. Evaluation

TL;DR

Model Details

Model Description

• Developed by: https://www.tii.ae
• Model type: Causal decoder-only
• Architecture: Mamba
• Language(s) (NLP): Mainly English
• License: TII Falcon-Mamba License 2.0

Check out the blogpost for more details!

Usage

Find below some example scripts on how to use the model in transformers (Make sure to have the latest transformers, or the one built from source):

Using the Pytorch model

Running the model on a CPU

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-mamba-7b")
model = AutoModelForCausalLM.from_pretrained("tiiuae/falcon-mamba-7b")

input_text = "Question: How many hours in one day? Answer: "
input_ids = tokenizer(input_text, return_tensors="pt").input_ids

outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))

Running the model on a GPU

# pip install accelerate
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-mamba-7b")
model = AutoModelForCausalLM.from_pretrained("tiiuae/falcon-mamba-7b", device_map="auto")

input_text = "Question: How many hours in one day? Answer: "
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")

outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))

Running the model on a GPU using torch.compile

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-mamba-7b")
model = AutoModelForCausalLM.from_pretrained("tiiuae/falcon-mamba-7b", torch_dtype=torch.bfloat16).to(0)

model = torch.compile(model)

input_text = "Question: How many hours in one day? Answer: "
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")

outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))

Running the model on a GPU using different precisions

FP16

# pip install accelerate
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-mamba-7b")
model = AutoModelForCausalLM.from_pretrained("tiiuae/falcon-mamba-7b", device_map="auto", torch_dtype=torch.float16)

input_text = "Question: How many hours in one day? Answer: "
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")

outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))

4-bit

# pip install bitsandbytes accelerate
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig

tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-mamba-7b")
model = AutoModelForCausalLM.from_pretrained("tiiuae/falcon-mamba-7b", device_map="auto", quantization_config=BitsAndBytesConfig(load_in_4bit=True))

input_text = "Question: How many hours in one day? Answer: "
input_ids = tokenizer(input_text, return_tensors="pt").input_ids.to("cuda")

outputs = model.generate(input_ids)
print(tokenizer.decode(outputs[0]))

Training Details

Training Data

Falcon-Mamba has been trained with ~ 5,500 GT mainly coming from Refined-Web, a large volume web-only dataset filtered and deduplicated. Similar to the others Falcon suite models, Falcon-Mamba has been trained leveraging a multi-stage training strategy to increase the context-length from 2,048 to 8,192. Moreover, inspired by the concept of Curriculum Learning, we carefully selected data mixtures throughout the training stages, considering both data diversity and complexity. Note that at inference the context-length is not relevant as the Mamba architecture has no limit on long range dependency. At the last training stage, small portion of high-quality curated data was used to further enhance performance.

Overall, the data sources included RefinedWeb-English, high quality technical data, code data and math data extracted from public sources. In particular, we used samples coming from Fineweb-edu during our last training stage.

The data was tokenized with the Falcon-7B/11B tokenizer.

Training Procedure

Falcon-Mamba-7B was trained on 256 H100 80GB GPUs for the majority of the training, using a 3D parallelism strategy (TP=1, PP=1, DP=256) combined with ZeRO.

Training Hyperparameters

The model was trained AdamW optimizer, WSD (warmup-stable-decay) learning rate schedule, and a batch size rampup from \(b_{\mathrm{min}}=128\) to \(b_{\mathrm{max}}=2048\) during first 50 GT of training. In the stable phase we used maximal learning rate \(\eta_{\mathrm{max}}=6.4 \times 10^{-4}\), and decayed it to the minimal value \(\eta_{\mathrm{min}}=\frac{\eta_{\mathrm{max}}}{256}\) with exponential schedule over 500 GT. Also, we applied BatchScaling during the rampup — rescaling learning rate \(\eta\) so that the Adam noise temperature \(T_{\mathrm{noise}}\