Hugging Face's logo

macrae
/
Llama-3-2-3B-rk3588

llama3.2
llama3.2-3B
rkllm
rockchip
rk3588
Model card Files Community
1
macrae commited on
Commit
037e2eb
·
verified ·
1 Parent(s): 6d617da

Update README.md

Browse files

add model with runtime 1.1.4

Files changed (1) hide show
  1. README.md +7 -2
README.md CHANGED
@@ -1,6 +1,5 @@
1
  ---
2
  license: llama3.2
3
- ---
4
  tags:
5
  - llama3.2
6
  - llama3.2-3B
@@ -12,10 +11,16 @@ tags:
12
  This is a conversion from https://huggingface.co/meta-llama/Meta-Llama-3.2-3B to the RKLLM format for Rockchip devices.
13
  Target conversion to be running on the RK3588 NPU.
14
 
15
- Model converted with **RKLLM runtime 1.1.1**, please follow this quantification to access the model:
16
  - [w8a8](https://huggingface.co/macrae/Llama-3-2-3B-rk3588/tree/w8a8-toolkitv1.1.1)
17
  - [w8a8_g512](https://huggingface.co/macrae/Llama-3-2-3B-rk3588/tree/w8a8_g512-toolkitv1.1.1)
18
 
 
 
 
 
 
 
19
  # License
20
  Same as the original LLM:
21
 
 
1
  ---
2
  license: llama3.2
 
3
  tags:
4
  - llama3.2
5
  - llama3.2-3B
 
11
  This is a conversion from https://huggingface.co/meta-llama/Meta-Llama-3.2-3B to the RKLLM format for Rockchip devices.
12
  Target conversion to be running on the RK3588 NPU.
13
 
14
+ **RKLLM runtime 1.1.1** requires **NPU driver 0.9.6**:
15
  - [w8a8](https://huggingface.co/macrae/Llama-3-2-3B-rk3588/tree/w8a8-toolkitv1.1.1)
16
  - [w8a8_g512](https://huggingface.co/macrae/Llama-3-2-3B-rk3588/tree/w8a8_g512-toolkitv1.1.1)
17
 
18
+ **RKLLM runtime 1.1.4** requires **NPU driver 0.9.8**:
19
+ - [w8a8](https://huggingface.co/macrae/Llama-3-2-3B-rk3588/tree/w8a8-toolkitv1.1.4)
20
+ - [w8a8_g512](https://huggingface.co/macrae/Llama-3-2-3B-rk3588//tree/w8a8_g512-toolkitv1.1.4)
21
+
22
+
23
+
24
  # License
25
  Same as the original LLM:
26