Gemma 1 2B
Parameters
2B
Context Length
8K
Modality
Text
Architecture
Dense
KV Cache Rate
9 KB/tok
License
Gemma Terms of Use
Release Date
21 Feb 2024
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.10 · Output: $0.10
Hardware Sizing
Precision
Context
Workload Memory: 3.50 GB
INT4 · 8K context
Base Weights: 0.93 GB
KV Cache: 0.07 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
3.5 / 21.6 GB usable
1x RTX 3060 12GB
3.5 / 10.8 GB usable
Apple M4 24GB
3.5 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
Rank
#240
| Benchmark | Score | Rank |
|---|---|---|
General Text | 1094 | 184 |
Rankings
Overall Rank
#240
Coding Rank
-
About Gemma 1 2B
Gemma 1 2B is a lightweight open language model by Google designed for on-device conversational tasks, summarization, and local NLP prototyping. It features Multi-Query Attention for low-latency inference on consumer-grade hardware.
Technical Specifications
Attention
Attention Structure
Multi-Query Attention
Attention Heads
16
Key-Value Heads
1
Position Embedding
ROPE
Normalization
RMS Normalization
Dimensions
Hidden Dimension Size
2,048
Number of Layers
18
About Gemma 1
Gemma 1 is a family of lightweight, decoder-only transformer models from Google, available in 2B and 7B parameter sizes. Designed for various text generation tasks, they incorporate rotary positional embeddings, shared input/output embeddings, GEGLU activation, and RMSNorm. The 2B model uses multi-query attention, while 7B uses multi-head attention.