File size: 1,503 Bytes
03e5196
c1c0791
 
 
 
 
 
 
 
 
 
 
 
03e5196
 
c1c0791
03e5196
c1c0791
 
 
 
03e5196
c1c0791
03e5196
c1c0791
 
 
03e5196
 
 
c1c0791
 
 
 
 
03e5196
c1c0791
03e5196
c1c0791
 
 
 
03e5196
c1c0791
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
---
language: en
license: apache-2.0
tags:
- pytorch
- jax-conversion
- transformers
- resnet
- hil-serl
- Lerobot
- vision
- image-classification
library_name: pytorch
---

# JAX to PyTorch Converted Model (ResNet-10)

It's done in context of porting `HIL-SERL` paper code (https://hil-serl.github.io/) to `Lerobot` (https://github.com/Lerobot/lerobot).
The HF doesn't have ResNet-10 model, which could be pretty usefult for robotics tasks because of it's small size.
This model is converted from JAX to PyTorch, and the weights are preserved.
## Model Description

[Brief description of the original model and its purpose]

This model is a PyTorch port of the original JAX implementation. The conversion maintains
the original model's architecture and weights while making it accessible to PyTorch users.
The original model is from https://github.com/rail-berkeley/hil-serl/blob/7d17d13560d85abffbd45facec17c4f9189c29c0/serl_launcher/serl_launcher/utils/train_utils.py#L103.

## Model Details

- **Original Framework:** JAX
- **Target Framework:** PyTorch
- **Model Architecture:** [Specify architecture]
- **Original Model:** [Link to original model]
- **Parameters:** [Number of parameters]

## Conversion Process

This model was converted using an automated JAX to PyTorch conversion pipeline, ensuring:
- Weight preservation
- Architecture matching
- Numerical stability

## Usage
```python
from transformers import AutoModel, AutoTokenizer
model = AutoModel.from_pretrained("lilkm/resnet10")
```