Falcon-7B
Parameters
7B
Context Length
2K
Modality
Text
Architecture
Dense
KV Cache Rate
8 KB/tok
License
Apache 2.0
Release Date
5 Jun 2023
Knowledge Cutoff
-
API Pricing (per 1M)
Self-hosted only
Hardware & System Requirements
Precision
Context
Workload Memory: 5.78 GB
INT4 · 2048 context
Base Weights: 3.26 GB
KV Cache: 0.02 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
5.8 / 21.6 GB usable
1x RTX 3060 12GB
5.8 / 10.8 GB usable
Apple M4 24GB
5.8 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Runtime requirements
Checkpoint format
FP16
CUDA
>= 11.8
PyTorch
>= 2.1
Transformers
>= 4.27.4
| Architecture | Precision | Attention kernel |
|---|---|---|
Blackwell, Hopper | Native FP16 | FlashAttention-3 |
Ada Lovelace, Ampere | Native FP16 | FlashAttention-2 |
Architecture Diagram
Evaluation Benchmarks
No evaluation benchmarks for Falcon-7B available.
Rankings
Overall Rank
-
Coding Rank
-
About Falcon-7B
Falcon-7B is an open-source 7B parameter causal language model developed by TII for text generation, conversational AI, and summarization. Trained on 1.5T tokens with Multi-Query Attention, it provides fast, accessible inference under Apache 2.0.
Specifications
Attention
Attention Structure
Multi-Query Attention
Attention Heads
71
Key-Value Heads
1
Position Embedding
ROPE
Normalization
Layer Normalization
Dimensions
Base Precision
FP16
Hidden Dimension Size
4,544
Number of Layers
32
Tokenizer
Vocabulary Size
65,024
About Falcon
The TII Falcon model family comprises causal decoder-only language models (7B, 40B). Their architecture, adapted from GPT-3, integrates rotary positional embeddings, Multi-Query Attention for inference efficiency, and FlashAttention for accelerated operations. Models are trained on the RefinedWeb dataset.