Update README.md
Browse files
README.md
CHANGED
@@ -94,13 +94,7 @@ This dataset was adapted for Albanian language training to ensure the model can
|
|
94 |
|
95 |
## Limitations
|
96 |
|
97 |
-
The current model is an 8-bit quantized version of the 27B parameter model.
|
98 |
-
|
99 |
-
- Reduced precision compared to the original 16-bit or 32-bit model
|
100 |
-
- May exhibit occasional numerical instabilities in complex reasoning chains
|
101 |
-
- While optimized for logical reasoning in Albanian, complex or ambiguous problems may produce inconsistent results
|
102 |
-
- As with all language models, it may occasionally hallucinate or provide incorrect information
|
103 |
-
- Performance may vary depending on the complexity and clarity of the input prompts
|
104 |
|
105 |
## Acknowledgments
|
106 |
- Google for developing the Gemma 3 architecture
|
|
|
94 |
|
95 |
## Limitations
|
96 |
|
97 |
+
The current model is an 8-bit quantized version of the 27B parameter model. It can run on much lower specifications, but at the cost of some performance.
|
|
|
|
|
|
|
|
|
|
|
|
|
98 |
|
99 |
## Acknowledgments
|
100 |
- Google for developing the Gemma 3 architecture
|