File size: 385 Bytes
ff3c5ea
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
---

datasets:
- openlamm/LAMM_Dataset
language:
- en

---
Repo:
- https://github.com/OpenLAMM/LAMM
- Commit ID: 6b98b26b3340378bb5bed330cca86305759d8d22

Model Config:
- LLM: Vicuna_13b_v0
- Vision Encoder: CLIP ViT-L-14
- lora_r: 32
- lora_alpha: 32
- lora_dropout: 0.1
- lora_target_modules: ['q_proj', 'k_proj', 'v_proj', 'o_proj']

Train Config:
- Epochs: 2
- train_batch_size: 64