v0.32.0
Browse filesSee https://github.com/quic/ai-hub-models/releases/v0.32.0 for changelog.
LICENSE
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
The license of the original trained model can be found at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct/blob/main/LICENSE.txt.
|
| 2 |
+
The license for the deployable model files (.tflite, .onnx, .dlc, .bin, etc.) can be found at https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct/blob/main/LICENSE.txt.
|
README.md
CHANGED
|
@@ -28,8 +28,6 @@ This model is an implementation of Llama-v3.2-3B-Instruct found [here](https://h
|
|
| 28 |
- **Model Stats:**
|
| 29 |
- Input sequence length for Prompt Processor: 128
|
| 30 |
- Context length: 4096
|
| 31 |
-
- Number of parameters: 3B
|
| 32 |
-
- Model size: 2.4G
|
| 33 |
- Precision: w4a16 + w8a16 (few layers)
|
| 34 |
- Num of key-value heads: 8
|
| 35 |
- Model-1 (Prompt Processor): Llama-PromptProcessor-Quantized
|
|
|
|
| 28 |
- **Model Stats:**
|
| 29 |
- Input sequence length for Prompt Processor: 128
|
| 30 |
- Context length: 4096
|
|
|
|
|
|
|
| 31 |
- Precision: w4a16 + w8a16 (few layers)
|
| 32 |
- Num of key-value heads: 8
|
| 33 |
- Model-1 (Prompt Processor): Llama-PromptProcessor-Quantized
|