flint-oQ6e-fp16
This model was quantized using oQ (oMLX v0.7.3rc1) mixed-precision quantization.
The effective average is 6.600 bits/weight for the complete checkpoint.
Quantization and Bit Distribution
| Storage format | Logical weights | Share of total | Tensor storage | Effective bits/weight |
|---|---|---|---|---|
| Affine 6-bit, group 64 | 37.849B | 95.134% | 28.641 GiB | 6.50 |
| Affine 8-bit, group 128 | 0.296B | 0.743% | 0.284 GiB | 8.25 |
| Affine 8-bit, group 64 | 1.615B | 4.060% | 1.598 GiB | 8.50 |
| FP16 | 0.025B | 0.063% | 0.046 GiB | 16.00 |
| Total | 39.785B | 100% | 30.569 GiB | 6.600 |
Storage Breakdown
| Component | Logical weights | Storage |
|---|---|---|
| Language model | 39.785B | 32.824 GB / 30.569 GiB |
| Total | 39.785B | 32.824 GB / 30.569 GiB |
- Downloads last month
- 72
Model size
40B params
Tensor type
U32
·
F16 ·
Hardware compatibility
Log In to add your hardware
6-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support