Falcon-3B
Parameters
3B
Context Length
33K
Modality
Text
Architecture
Dense
KV Cache Rate
4 KB/tok
License
TII Falcon-LLM License 2.0
Release Date
17 Dec 2024
Knowledge Cutoff
-
API Pricing (per 1M)
Self-hosted only
Hardware & System Requirements
Precision
Context
Workload Memory: 3.93 GB
INT4 · 8K context
Base Weights: 1.40 GB
KV Cache: 0.03 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
3.9 / 21.6 GB usable
1x RTX 3060 12GB
3.9 / 10.8 GB usable
Apple M4 24GB
3.9 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
No evaluation benchmarks for Falcon-3B available.
Rankings
Overall Rank
-
Coding Rank
-
About Falcon-3B
Falcon-3B is a compact 3B parameter open foundation model from TII distilled for reasoning, code generation, and instruction following. It supports multilingual capabilities across English, French, Spanish, and Portuguese over a 32K context window.
Specifications
Attention
Attention Structure
Multi-Query Attention
Attention Heads
48
Key-Value Heads
1
Position Embedding
ROPE
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Base Precision
BF16
Hidden Dimension Size
1,536
Number of Layers
32
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.