xing/amlnn-model-playground
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update README.md

Browse source
This commit is contained in:
yuan dian 2026-01-30 16:20:20 +08:00 • committed by GitHub
parent 4e2f827908
commit 4bf4aafc73
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
examples/LLMs/README.md
View file

@ -39,7 +39,7 @@ ADLA2: A311D2_3.2T / S905X5_4T
Pre-quantized ADLA models are available on Hugging Face: Pre-quantized ADLA models are available on Hugging Face:
- **Qwen2.5-1.5B (A311D2)**: [Hugging Face Repository](https://huggingface.co/Amlogic-NN/Qwen2.5-1.5B-Instruct-F16_quant_i8_adla/tree/main/a311d2) - **Qwen2.5-0.5B (A311D2)**: [Hugging Face Repository](https://huggingface.co/Amlogic-NN/Qwen2.5-0.5B-Instruct_quant_i8/blob/main/Qwen2.5-0.5B-Instruct_quant_i8_a311d2.adla)
## Compile ## Compile
@ -87,7 +87,7 @@ To compile the CPP project using Android NDK, follow these steps:
Then execute: Then execute:
```bash ```bash
./demo_llm_main Qwen2.5-1.5B-Instruct-F16_quant_i8_t7c.adla tokenizer.json ./demo_llm_main Qwen2.5-0.5B-Instruct_quant_i8_a311d2.adla tokenizer.json
``` ```
### Python ### Python
@ -131,7 +131,7 @@ The driver version must be 1.7.x or higher.
4. **Usage Examples**: 4. **Usage Examples**:
```bash ```bash
# Using Qwen model # Using Qwen model
python simple_chat.py --model Qwen2.5-1.5B-Instruct-F16_quant_i8_t7c.adla --tokenizer tokenizer.json --model-type qwen python simple_chat.py --model Qwen2.5-0.5B-Instruct_quant_i8_a311d2.adla --tokenizer tokenizer.json --model-type qwen
# Using Top-P sampling mode # Using Top-P sampling mode
python simple_chat.py --model model.adla --tokenizer tokenizer.json --sampling-mode top_p --top-p 0.9 --temperature 0.8 python simple_chat.py --model model.adla --tokenizer tokenizer.json --sampling-mode top_p --top-p 0.9 --temperature 0.8
@ -152,4 +152,4 @@ The driver version must be 1.7.x or higher.
| Banner | Inference Result | | Banner | Inference Result |
| :---: | :---: | | :---: | :---: |
| ![llm-result0](./model/llm-result0.png) | ![llm-result](./model/llm_result.png) | | ![llm-result0](./model/llm-result0.png) | ![llm-result](./model/llm_result.png) |