EmbeddingGemma 2 oQ
Collection
18 items • Updated
embeddinggemma-2-oQ2.5
This model was quantized using oQ (oMLX v0.7.1) mixed-precision quantization.
The effective average is 11.651 bits/weight for the complete checkpoint.
Quantization and Bit Distribution
| Storage format | Logical weights | Share of total | Tensor storage | Effective bits/weight |
|---|---|---|---|---|
| Affine 2-bit, group 64 | 0.142B | 19.088% | 0.041 GiB | 2.50 |
| Affine 3-bit, group 64 | 0.045B | 6.057% | 0.018 GiB | 3.50 |
| Affine 4-bit, group 64 | 0.016B | 2.148% | 0.008 GiB | 4.50 |
| Affine 6-bit, group 64 | 0.027B | 3.663% | 0.021 GiB | 6.50 |
| Affine 8-bit, group 64 | 0.042B | 5.599% | 0.041 GiB | 8.50 |
| BF16 | 0.472B | 63.445% | 0.880 GiB | 16.00 |
| Total | 0.744B | 100% | 1.010 GiB | 11.651 |
Storage Breakdown
| Component | Logical weights | Storage |
|---|---|---|
| Language model | 0.272B | 0.140 GB / 0.130 GiB |
| Vision | 0.167B | 0.335 GB / 0.312 GiB |
| Audio | 0.305B | 0.610 GB / 0.568 GiB |
| Total | 0.744B | 1.084 GB / 1.010 GiB |
- Downloads last month
- 23
Model size
0.7B params
Tensor type
BF16
·
U32 ·
Hardware compatibility
Log In to add your hardware
2-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support