A1-Q6_K-imatrix
Static imatrix-calibrated GGUF quant of InternScience/Agents-A1.
llama-cli -hf Chungulus/Agents-A1-Q6_K-imatrix-gguf-fable5-calibrated:Agents-A1-Q6_K-imatrix-gguf-fable5-calibrated.gguf -p "Write a Python sorting function" -n 160
File
| File | Size | SHA-256 |
|---|---|---|
Agents-A1-Q6_K-imatrix-gguf-fable5-calibrated.gguf |
26.56 GiB | 4ddb5fc1a537d364391a4adc4172d364ec31782e6b1e4d802e8fbe7b02e99ee9 |
Quality Snapshot
F16 baseline mini accuracy: 89.58%. F16 baseline PPL on KL holdout: 13.0194.
| Metric | Value |
|---|---|
| Mini accuracy | 89.58% |
| Retention vs F16 | 100.00% |
| Mean KLD vs F16 | 0.007926 |
| Same top p | 94.44% |
Notes
- Calibration source:
Glint-Research/Fable-5-traces - MTP is not included because the downloaded checkpoint did not contain MTP tensors.
- Static imatrix GGUF, not Unsloth Dynamic 2.0 / UD2.