docs: remove AWQ variants (AWQ lane paused; MLX focus)
Browse files
README.md
CHANGED
|
@@ -35,7 +35,6 @@ KV cache compression reduces the memory used by the attention cache during infer
|
|
| 35 |
|
| 36 |
| Technique | Where it's applied | Savings |
|
| 37 |
|-----------|-------------------|---------|
|
| 38 |
-
| Weight quantization (GGUF/MLX/AWQ) | Baked into model file | Reduces disk + weight memory |
|
| 39 |
| **RotorQuant KV cache** | At inference time | Reduces attention memory (critical for long context) |
|
| 40 |
|
| 41 |
Both can be combined for maximum efficiency.
|
|
|
|
| 35 |
|
| 36 |
| Technique | Where it's applied | Savings |
|
| 37 |
|-----------|-------------------|---------|
|
|
|
|
| 38 |
| **RotorQuant KV cache** | At inference time | Reduces attention memory (critical for long context) |
|
| 39 |
|
| 40 |
Both can be combined for maximum efficiency.
|