ZJU-Fangyin
commited on
Commit
•
19c153d
1
Parent(s):
5948b3f
Update README.md
Browse files
README.md
CHANGED
@@ -87,6 +87,26 @@ Instructions for running it can be found at https://github.com/zjunlp/Mol-Instru
|
|
87 |
</details>
|
88 |
|
89 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
90 |
<h3> 🚨 Limitations</h3>
|
91 |
|
92 |
The current state of the model, obtained via instruction tuning, is a preliminary demonstration. Its capacity to handle real-world, production-grade tasks remains limited.
|
|
|
87 |
</details>
|
88 |
|
89 |
|
90 |
+
<h3> 📝 Demo</h3>
|
91 |
+
|
92 |
+
As illustrated in [our repository](https://github.com/zjunlp/Mol-Instructions/tree/main/demo), we provide an example to perform generation.
|
93 |
+
|
94 |
+
```shell
|
95 |
+
>> python generate.py \
|
96 |
+
--CLI True \
|
97 |
+
--protein False\
|
98 |
+
--load_8bit \
|
99 |
+
--base_model $BASE_MODEL_PATH \
|
100 |
+
--lora_weights $FINETUNED_MODEL_PATH \
|
101 |
+
```
|
102 |
+
|
103 |
+
Please download [llama-7b-hf](https://huggingface.co/decapoda-research/llama-7b-hf/tree/main) to obtain the pre-training weights of LLaMA-7B, refine the `--base_model` to point towards the location where the model weights are saved.
|
104 |
+
|
105 |
+
For model fine-tuned on **molecule-oriented** instructions, set `$FINETUNED_MODEL_PATH` to `'zjunlp/llama-molinst-molecule-7b'`.
|
106 |
+
|
107 |
+
|
108 |
+
|
109 |
+
|
110 |
<h3> 🚨 Limitations</h3>
|
111 |
|
112 |
The current state of the model, obtained via instruction tuning, is a preliminary demonstration. Its capacity to handle real-world, production-grade tasks remains limited.
|