Add note about dev effort and GH shadowban
Browse files
README.md
CHANGED
|
@@ -51,6 +51,8 @@ ASHQ1 at 5700 MiB beats uniform Q6_K by 0.047 PPL at 19% smaller size.
|
|
| 51 |
|
| 52 |
> Note: The model file is being uploaded now. My internet connection is very slow (~100 KB/s), so a full 5.6 GB upload takes around 20 hours. If the GGUF file is not yet available in the repo, it is still being uploaded. Please be patient.
|
| 53 |
|
|
|
|
|
|
|
| 54 |
## How It Works
|
| 55 |
|
| 56 |
The classifier operates in a single-pass max-heap to drain a strictly defined size budget.
|
|
|
|
| 51 |
|
| 52 |
> Note: The model file is being uploaded now. My internet connection is very slow (~100 KB/s), so a full 5.6 GB upload takes around 20 hours. If the GGUF file is not yet available in the repo, it is still being uploaded. Please be patient.
|
| 53 |
|
| 54 |
+
> I have put a lot of effort into developing this quantization method. ASHQ1 may not be released as open-source on GitHub due to a shadowban on my account and the difficulty of maintaining the project. This HF repo is the primary distribution channel.
|
| 55 |
+
|
| 56 |
## How It Works
|
| 57 |
|
| 58 |
The classifier operates in a single-pass max-heap to drain a strictly defined size budget.
|