Update README.md
Browse files
README.md
CHANGED
|
@@ -46,18 +46,20 @@ tiny/
|
|
| 46 |
|
| 47 |
## Available Formats
|
| 48 |
|
|
|
|
|
|
|
| 49 |
| Format | Quality | Use Case |
|
| 50 |
|--------| ---------|----------|
|
| 51 |
| q2_k | Smallest | Extreme compression |
|
| 52 |
| q3_k | Small | Mobile devices |
|
| 53 |
-
| q4_0 |
|
| 54 |
-
| q4_k |
|
| 55 |
-
| q4_1 |
|
| 56 |
-
| q5_0 |
|
| 57 |
-
| q5_k |
|
| 58 |
-
| q5_1 | Very Good
|
| 59 |
| q6_k | Excellent | Near-lossless |
|
| 60 |
-
| q8_0 | Excellent | Minimal loss, benchmarking |
|
| 61 |
|
| 62 |
## Usage
|
| 63 |
|
|
|
|
| 46 |
|
| 47 |
## Available Formats
|
| 48 |
|
| 49 |
+
**Q5_0** is the "last good" option. Lower quantizations (Q4, Q3, Q2) lead to a sharp loss of quality and meaningless output.
|
| 50 |
+
|
| 51 |
| Format | Quality | Use Case |
|
| 52 |
|--------| ---------|----------|
|
| 53 |
| q2_k | Smallest | Extreme compression |
|
| 54 |
| q3_k | Small | Mobile devices |
|
| 55 |
+
| q4_0 | Small | Legacy compatibility |
|
| 56 |
+
| q4_k | Small | - |
|
| 57 |
+
| q4_1 | Small | Legacy with bias |
|
| 58 |
+
| q5_0 | Good | Legacy compatibility |
|
| 59 |
+
| q5_k | Good | Good quality |
|
| 60 |
+
| q5_1 | Very Good | Legacy with bias |
|
| 61 |
| q6_k | Excellent | Near-lossless |
|
| 62 |
+
| q8_0 | Excellent | **Recommended for production** Minimal loss, benchmarking |
|
| 63 |
|
| 64 |
## Usage
|
| 65 |
|