Gemma 2 9B
Parameters
9B
Context Length
8K
Modality
Text
Architecture
Dense
KV Cache Rate
95 KB/tok
License
Gemma License
Release Date
27 Jun 2024
Knowledge Cutoff
-
API Pricing (per 1M)
Input: $0.03 · Output: $0.09
Hardware Sizing
Precision
Context
Workload Memory: 7.43 GB
INT4 · 8K context
Base Weights: 4.19 GB
KV Cache: 0.74 GB
Runtime & Graphs: 2.50 GB
Minimum Required Hardware
1x NVIDIA L4 24GB
7.4 / 21.6 GB usable
1x RTX 3060 12GB
7.4 / 10.8 GB usable
Apple M4 24GB
7.4 / 18.0 GB usable
Determine requirements for your specific workload:
Open in VRAM Calculator
Architecture Diagram
Evaluation Benchmarks
Rank
#207
| Benchmark | Score | Rank |
|---|---|---|
General Text | 1267 | 164 |
StackEval Archived | 0.715 | 16 |
QA Assistant Archived | 0.822 | 18 |
General Knowledge Reference | 0.713 | 24 |
Summarization Archived | 0.582 | 26 |
Rankings
Overall Rank
#207
Coding Rank
-
About Gemma 2 9B
Gemma 2 9B is a versatile open language model by Google engineered for conversational AI, text generation, and code assistance. Trained on 8T tokens with knowledge distillation, it delivers strong performance across an 8K context window.
Technical Specifications
Attention
Attention Structure
Grouped-Query Attention
Attention Heads
32
Key-Value Heads
8
Position Embedding
ROPE
Normalization
RMS Normalization
Activation Function
SwigLU
Dimensions
Hidden Dimension Size
2,304
Number of Layers
42
About Gemma 2
Gemma 2 is Google's family of open large language models, offering 2B, 9B, and 27B parameter sizes. Built upon the Gemma architecture, it incorporates innovations such as interleaved local and global attention, logit soft-capping for training stability, and Grouped Query Attention for inference efficiency. The smaller models leverage knowledge distillation.