Gemma 2 2B IT: memory fit on every Jetson module.
Method v1.0 · dataset 2026-09-07 · verified 2026-09-07 · last updated September 2026
2.61 B parameters · 26 layers · 8,192-token max context. Hugging Face: google/gemma-2-2b-it.
Model facts
| Fact | Value | Note |
|---|---|---|
| Parameters | 2.614 B | |
| Layers | 26 | |
| Hidden size | 2304 | |
| Attention heads | 8 | |
| KV heads | 4 | |
| Head dimension | 256 | |
| Max context | 8,192 tokens | |
| dtype | bfloat16 | |
| Vocabulary | 256,000 |
Hugging Face (mirror: unsloth/gemma-2-2b-it) · hf:unsloth/gemma-2-2b-it · config.json · verified 2026-09-07
“"num_attention_heads": 8, "num_key_value_heads": 4, "head_dim": 256”
Parameters source
- Hugging Face (mirror: unsloth/gemma-2-2b-it, identical weights to google/gemma-2-2b-it) · hf:unsloth/gemma-2-2b-it · verified 2026-09-07 · model.safetensors (blob metadata via HF API) · class A
“model.safetensors size 5228717512 bytes”
Published quantised artefacts
A file the publisher or a community mirror actually ships, used as the class-A weight figure when the requested quant matches. Any quant without a row here falls back to parameters × bytes-per-parameter (class D for block quants).
| Quant | Format | File | Size | Source |
|---|---|---|---|---|
| GGUF Q4_K_M / INT4 (AWQ, MLC q4f16) | GGUF Q4_K_M | gemma-2-2b-it-Q4_K_M.gguf | 1.709 GB | bartowski (community GGUF) · hf:bartowski/gemma-2-2b-it-GGUF · verified 2026-09-07 |
Jetson tokens/s measurements
Published or archived throughput numbers, not modelled. Class C (external measured benchmark).
| Module | Runtime | Quant | Measured | Source |
|---|---|---|---|---|
| Jetson Orin Nano Super | unspecified (archived benchmark) | — | 34.97 tok/s | NVIDIA Jetson AI Lab (archive) · jetson-ai-lab:benchmarks.html · verified 2026-09-07 |
Every module × every quantisation
Verdict and total memory at 4096-token context, 1 sequence, llama.cpp, headless. Q4 cells link to the static breakdown page; every other cell links to the live engine at that quantisation.
Notes
google/gemma-2-2b-it config.json/README return 401 (gated); unsloth mirror (byte-identical weights) used.
Method and data: Model Memory Fit methodology. Full registry: model-memory.json.
Change the context, concurrency or runtime.
The live engine covers any context length, sequence count, KV precision and runtime, with a permanent link.