update hf weight link

README.MD

@@ -49,28 +49,28 @@ For the detailed training configuration, please refer to the [t5x branch](https:
 <td align="center">8640</td>
 <td align="center">$13613</td>
 <td align="center">79.1</td>
-<td align="center"><a href="https://drive.google.com/file/d/1EiQpLvL51AXEFzJ33e6z58N0dQ83CSux/view?usp=sharing">PyTorch / <a href="https://drive.google.com/file/d/1IQ0BgWGy0Tsui9iK_wdVifFKc3NfC0BD/view?usp=sharing">JAX</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-H-14-CLIPA-336-laion2B">PyTorch / <a href="https://drive.google.com/file/d/1IQ0BgWGy0Tsui9iK_wdVifFKc3NfC0BD/view?usp=sharing">JAX</td>
 <tr><td align="left">L/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224 +128M@336 </td>
 <td align="center">4520</td>
 <td align="center">$7124</td>
 <td align="center">80.3</td>
-<td align="center"><a href="https://drive.google.com/file/d/1Vpon6Dn0E3xDfyCIuOW1SPo9haKYvFiD/view?usp=sharing">PyTorch / <a href="https://drive.google.com/file/d/1yJD8p27HdZGK2DZ8x64uR6idl_QNkL7w/view?usp=sharing">JAX</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-L-14-CLIPA-336-datacomp1B">PyTorch / <a href="https://drive.google.com/file/d/1yJD8p27HdZGK2DZ8x64uR6idl_QNkL7w/view?usp=sharing">JAX</td>
 <tr><td align="left">H/14</td>
 <td align="center">DataCOMP-1B</td>
-<td align="center">12.8B@84 + 512M@224 + +128M@336</td>
+<td align="center">12.8B@84 + 512M@224 + 128M@336</td>
 <td align="center">8640</td>
 <td align="center">$13613</td>
 <td align="center">81.8</td>
-<td align="center"><a href="https://drive.google.com/file/d/1oOACMg3MKXUpG-xn-UrqDWFVEIvenA-F/view?usp=sharing">PyTorch / <a href="https://drive.google.com/file/d/1t0k_5m3VVLRyThUjc4JQ_-z29hABuFBV/view?usp=sharing">JAX</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-H-14-CLIPA-336-datacomp1B">PyTorch / <a href="https://drive.google.com/file/d/1t0k_5m3VVLRyThUjc4JQ_-z29hABuFBV/view?usp=sharing">JAX</td>
 <tr><td align="left">bigG/14</td>
 <td align="center">DataCOMP-1B</td>
-<td align="center">12.8B@84 + 512M@224 + +128M@336</td>
+<td align="center">12.8B@84 + 512M@224 + 128M@336</td>
 <td align="center">23742</td>
 <td align="center">$39056</td>
 <td align="center">83.0</td>
-<td align="center"><a href="https://drive.google.com/file/d/1WpSMISoitcrBjQr2i9jXOVc7fJGEGFZc/view?usp=drive_link">PyTorch / <a href="https://drive.google.com/file/d/1WwXPYb61Zoz-Zd3R5sRwEIGlPRDj3QJo/view?usp=sharing">JAX</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-bigG-14-CLIPA-336-datacomp1B">PyTorch / <a href="https://drive.google.com/file/d/1WwXPYb61Zoz-Zd3R5sRwEIGlPRDj3QJo/view?usp=sharing">JAX</td>
 </tbody></table>
 
 Our CLIPA-v2’s GPU hour is estimated using an 8-A100 80GB GPU machine on Google Cloud. 

clipa_torch/README.md

@@ -128,21 +128,21 @@ Try our example testing scripts for evaluation.
 <td align="center">8640</td>
 <td align="center">$13616</td>
 <td align="center">79.1</td>
-<td align="center"><a href="https://drive.google.com/file/d/1EiQpLvL51AXEFzJ33e6z58N0dQ83CSux/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-H-14-CLIPA-336-laion2B">download</td>
 <tr><td align="left">L/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224</td>
 <td align="center">4008</td>
 <td align="center">$6318</td>
 <td align="center">79.7</td>
-<td align="center"><a href="https://drive.google.com/file/d/1PZCZZ-mxHnye_fluCPxqHSdm5SmF9BCT/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-L-14-CLIPA-datacomp1B">download</td>
 <tr><td align="left">L/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224 + 128M@336</td>
 <td align="center">4520</td>
 <td align="center">$7124</td>
 <td align="center">80.3</td>
-<td align="center"><a href="https://drive.google.com/file/d/1Vpon6Dn0E3xDfyCIuOW1SPo9haKYvFiD/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-L-14-CLIPA-336-datacomp1B">download</td>
 <tr><td align="left">H/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@70 + 512M@224</td>
@@ -156,28 +156,28 @@ Try our example testing scripts for evaluation.
 <td align="center">7776</td>
 <td align="center">$12247</td>
 <td align="center">81.5</td>
-<td align="center"><a href="https://drive.google.com/file/d/1JwnpWGgMV29svZRTZR8gPm_2ieZcPAy6/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-H-14-CLIPA-datacomp1B">download</td>
 <tr><td align="left">H/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224 + 128M@336</td>
 <td align="center">8640</td>
 <td align="center">$13616</td>
 <td align="center">81.8</td>
-<td align="center"><a href="https://drive.google.com/file/d/1oOACMg3MKXUpG-xn-UrqDWFVEIvenA-F/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-H-14-CLIPA-336-datacomp1B">download</td>
 <tr><td align="left">bigG/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224</td>
 <td align="center">21998</td>
 <td align="center">$34646</td>
 <td align="center">82.7</td>
-<td align="center"><a href="https://drive.google.com/file/d/1M-Zu-RddUIX7NTUQwUw_8LvXFNNoAhgk/view?usp=sharing">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-bigG-14-CLIPA-datacomp1B">download</td>
 <tr><td align="left">bigG/14</td>
 <td align="center">DataCOMP-1B</td>
 <td align="center">12.8B@84 + 512M@224 + 128M@336</td>
 <td align="center">23742</td>
 <td align="center">$39056</td>
 <td align="center">83.0</td>
-<td align="center"><a href="https://drive.google.com/file/d/1WpSMISoitcrBjQr2i9jXOVc7fJGEGFZc/view?usp=drive_link">download</td>
+<td align="center"><a href="https://huggingface.co/UCSC-VLAA/ViT-bigG-14-CLIPA-336-datacomp1B">download</td>
 </tbody></table>
 
 Our CLIPA-v2’s GPU hour is estimated using an 8-A100 80GB GPU machine on Google Cloud.