MiniGPT-Med

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 MiniGPT-Med-github/Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg filter=lfs diff=lfs merge=lfs -text
+Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,13 +1,51 @@
----
-title: MiniGPT Med
-emoji: 🌍
-colorFrom: pink
-colorTo: green
-sdk: gradio
-sdk_version: 4.41.0
-app_file: app.py
-pinned: false
-license: mit
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# MiniGPT-Med: Large Language Model as a General Interface for Radiology Diagnosis
+Asma Alkhaldi, Raneem Alnajim, Layan Alabdullatef, Rawan Alyahya, Jun Chen, Deyao Zhu, Ahmed Alsinan, Mohamed Elhoseiny
+
+*Saudi Data and Artificial Intelligence Authority (SDAIA) and King Abdullah University of Science and Technology (KAUST)*
+
+## Installation
+```
+git clone https://github.com/Vision-CAIR/MiniGPT-Med
+cd MiniGPT-Med
+conda env create -f environment.yml
+conda activate miniGPT-Med
+```
+
+## Download miniGPT-Med trained model weights
+
+* miniGPT-Med's weights [miniGPT-Med Model](https://drive.google.com/file/d/1kjGLk6s9LsBmXfLWQFCdlwF3aul08Cl8/view?usp=sharing)
+
+* Then modify line 8 at miniGPT-Med/eval_configs/minigptv2_eval.yaml to be the path of miniGPT-Med weight.
+
+## Prepare weight for LLMs
+
+### Llama2 Version
+
+```shell
+git clone https://huggingface.co/meta-llama/Llama-2-13b-chat-hf
+```
+
+Then modify line 14 at miniGPT-Med/minigpt4/configs/models/minigpt_v2.yaml to be the path of Llama-2-13b-chat-hf.
+
+## Launching Demo Locally
+
+```
+python demo.py --cfg-path eval_configs/minigptv2_eval.yaml --gpu-id 0
+```
+
+## Dataset
+| Dataset | Images  | json file| 
+|---------|---------|----------|
+| MIMIC   |[Download](https://physionet.org/content/mimiciii/1.4/) | [Download](https://drive.google.com/drive/folders/1nZhdfNoh7fkx7CWvf0_47_OLv3tA2m3o?usp=sharing) |
+| NLST    |[Download](https://wiki.cancerimagingarchive.net/display/NLST)| [Downlaod](https://drive.google.com/drive/folders/1OKgMTaGLu_dWRuco6JipYzezw3oNwgaz?usp=sharing) |
+|SLAKE    |[Downlaod](https://www.med-vqa.com/slake/) |[Download](https://drive.google.com/drive/folders/1vstjmfRbKahSAsi_b6FmTQiuolvgO8oC?usp=sharing)|
+|RSNA     |[Downlaod](https://www.rsna.org/rsnai/ai-image-challenge/rsna-pneumonia-detection-challenge-2018) | [Download](https://drive.google.com/drive/folders/1wkXPvUNqda6jWAIduyiVJkS3Tx7P7td8?usp=sharing) |
+|Rad-VQA  |[Downalod](https://osf.io/89kps/) |[Download](https://drive.google.com/drive/folders/1ING6Dodwk2DU_t4GHQYudNFMMg9OMfBQ?usp=sharing) |
+
+## Acknowledgement
+
+- MiniGPT-4
+- Lavis
+- Vicuna
+- Falcon
+- Llama 2

dcgm/bash/34649895/dcgm-gpu-stats-gpu202-02-r-34649895.out

@@ -0,0 +1,39 @@
+Successfully retrieved statistics for job: 34649895. 
++------------------------------------------------------------------------------+
+| GPU ID: 0                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Tue Jul  9 09:29:46 2024                |
+| End Time                           | Wed Jul 10 09:30:32 2024                |
+| Total Execution Time (sec)         | 86445.3                                 |
+| No. of Processes                   | 1                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 232291                                  |
+| Power Usage (Watts)                | Avg: 65.6704, Max: 84.315, Min: 61.555  |
+| Max GPU Memory Used (bytes)        | 10104078336                             |
+| SM Clock (MHz)                     | Avg: 595, Max: 1155, Min: 210           |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 1548651                                 |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34673507/dcgm-gpu-stats-gpu201-23-l-34673507.out

@@ -0,0 +1,39 @@
+Successfully retrieved statistics for job: 34673507. 
++------------------------------------------------------------------------------+
+| GPU ID: 1                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Fri Jul 12 11:48:45 2024                |
+| End Time                           | Sat Jul 13 11:49:39 2024                |
+| Total Execution Time (sec)         | 86454.5                                 |
+| No. of Processes                   | 1                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 252136                                  |
+| Power Usage (Watts)                | Avg: 69.7762, Max: 70.022, Min: 69.151  |
+| Max GPU Memory Used (bytes)        | 10104078336                             |
+| SM Clock (MHz)                     | Avg: 1157, Max: 1410, Min: 1155         |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 2527521                                 |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34676162/dcgm-gpu-stats-gpu201-23-l-34676162.out

@@ -0,0 +1,39 @@
+Successfully retrieved statistics for job: 34676162. 
++------------------------------------------------------------------------------+
+| GPU ID: 3                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Sun Jul 14 07:57:08 2024                |
+| End Time                           | Mon Jul 15 07:57:59 2024                |
+| Total Execution Time (sec)         | 86450.6                                 |
+| No. of Processes                   | 1                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 249997                                  |
+| Power Usage (Watts)                | Avg: 82.8167, Max: 86.615, Min: 70.491  |
+| Max GPU Memory Used (bytes)        | 10104078336                             |
+| SM Clock (MHz)                     | Avg: 1352, Max: 1410, Min: 1080         |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 3048225                                 |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34691276/dcgm-gpu-stats-gpu201-09-l-34691276.out

@@ -0,0 +1,42 @@
+Successfully retrieved statistics for job: 34691276. 
++------------------------------------------------------------------------------+
+| GPU ID: 0                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Tue Jul 16 08:21:43 2024                |
+| End Time                           | Tue Jul 16 21:44:34 2024                |
+| Total Execution Time (sec)         | 48170.9                                 |
+| No. of Processes                   | 2                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 222759                                  |
+| Power Usage (Watts)                | Avg: 61.4158, Max: 61.683, Min: 61.349  |
+| Max GPU Memory Used (bytes)        | 10806624256                             |
+| SM Clock (MHz)                     | Avg: 210, Max: 225, Min: 210            |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 1958147                                 |
+|     Avg SM Utilization (%)         | 1                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 2068287                                 |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34709014/dcgm-gpu-stats-gpu109-16-l-34709014.out

@@ -0,0 +1,39 @@
+Successfully retrieved statistics for job: 34709014. 
++------------------------------------------------------------------------------+
+| GPU ID: 3                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Thu Jul 18 07:54:11 2024                |
+| End Time                           | Fri Jul 19 07:55:07 2024                |
+| Total Execution Time (sec)         | 86456.3                                 |
+| No. of Processes                   | 1                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 245376                                  |
+| Power Usage (Watts)                | Avg: 67.8347, Max: 68.156, Min: 67.563  |
+| Max GPU Memory Used (bytes)        | 10582228992                             |
+| SM Clock (MHz)                     | Avg: 1161, Max: 1410, Min: 1155         |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 4005887                                 |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34721198/dcgm-gpu-stats-gpu203-23-r-34721198.out

@@ -0,0 +1,57 @@
+Successfully retrieved statistics for job: 34721198. 
++------------------------------------------------------------------------------+
+| GPU ID: 1                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Fri Jul 19 21:34:44 2024                |
+| End Time                           | Sat Jul 20 00:01:06 2024                |
+| Total Execution Time (sec)         | 8782.23                                 |
+| No. of Processes                   | 7                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 225540                                  |
+| Power Usage (Watts)                | Avg: 75.9496, Max: 87.541, Min: 65.792  |
+| Max GPU Memory Used (bytes)        | 13356761088                             |
+| SM Clock (MHz)                     | Avg: 210, Max: 210, Min: 210            |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 866309                                  |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 866955                                  |
+|     Avg SM Utilization (%)         | 1                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 868076                                  |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 868638                                  |
+|     Avg SM Utilization (%)         | 5                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 869519                                  |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 871043                                  |
+|     Avg SM Utilization (%)         | 1                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 871322                                  |
+|     Avg SM Utilization (%)         | 0                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34734121/dcgm-gpu-stats-gpu201-23-l-34734121.out

@@ -0,0 +1,35 @@
+Successfully retrieved statistics for job: 34734121. 
++------------------------------------------------------------------------------+
+| GPU ID: 3                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Tue Jul 23 11:47:49 2024                |
+| End Time                           | Tue Jul 23 13:47:51 2024                |
+| Total Execution Time (sec)         | 7202.22                                 |
+| No. of Processes                   | 0                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 226384                                  |
+| Power Usage (Watts)                | Avg: 62.6807, Max: 81.445, Min: 62.015  |
+| Max GPU Memory Used (bytes)        | 0                                       |
+| SM Clock (MHz)                     | Avg: 220, Max: 1410, Min: 210           |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34738689/dcgm-gpu-stats-gpu201-16-r-34738689.out

@@ -0,0 +1,35 @@
+Successfully retrieved statistics for job: 34738689. 
++------------------------------------------------------------------------------+
+| GPU ID: 3                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Wed Jul 24 10:14:38 2024                |
+| End Time                           | Wed Jul 24 11:45:33 2024                |
+| Total Execution Time (sec)         | 5454.69                                 |
+| No. of Processes                   | 0                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 232516                                  |
+| Power Usage (Watts)                | Avg: 64.2532, Max: 64.329, Min: 63.938  |
+| Max GPU Memory Used (bytes)        | 0                                       |
+| SM Clock (MHz)                     | Avg: 210, Max: 210, Min: 210            |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

dcgm/bash/34757693/dcgm-gpu-stats-gpu202-16-r-34757693.out

@@ -0,0 +1,42 @@
+Successfully retrieved statistics for job: 34757693. 
++------------------------------------------------------------------------------+
+| GPU ID: 2                                                                    |
++====================================+=========================================+
+|-----  Execution Stats  ------------+-----------------------------------------|
+| Start Time                         | Thu Jul 25 15:38:16 2024                |
+| End Time                           | Thu Jul 25 17:08:59 2024                |
+| Total Execution Time (sec)         | 5442.54                                 |
+| No. of Processes                   | 2                                       |
++-----  Performance Stats  ----------+-----------------------------------------+
+| Energy Consumed (Joules)           | 214029                                  |
+| Power Usage (Watts)                | Avg: 59.2012, Max: 67.659, Min: 59.026  |
+| Max GPU Memory Used (bytes)        | 7616856064                              |
+| SM Clock (MHz)                     | Avg: 243, Max: 1080, Min: 210           |
+| Memory Clock (MHz)                 | Avg: 1593, Max: 1593, Min: 1593         |
+| SM Utilization (%)                 | Avg: 0, Max: 0, Min: 0                  |
+| Memory Utilization (%)             | Avg: 0, Max: 0, Min: 0                  |
+| PCIe Rx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
+| PCIe Tx Bandwidth (megabytes)      | Avg: N/A, Max: N/A, Min: N/A            |
++-----  Event Stats  ----------------+-----------------------------------------+
+| Single Bit ECC Errors              | 0                                       |
+| Double Bit ECC Errors              | 0                                       |
+| PCIe Replay Warnings               | 0                                       |
+| Critical XID Errors                | 0                                       |
++-----  Slowdown Stats  -------------+-----------------------------------------+
+| Due to - Power (%)                 | 0                                       |
+|        - Thermal (%)               | 0                                       |
+|        - Reliability (%)           | Not Supported                           |
+|        - Board Limit (%)           | Not Supported                           |
+|        - Low Utilization (%)       | Not Supported                           |
+|        - Sync Boost (%)            | 0                                       |
++--  Compute Process Utilization  ---+-----------------------------------------+
+| PID                                | 1095606                                 |
+|     Avg SM Utilization (%)         | 3                                       |
+|     Avg Memory Utilization (%)     | 0                                       |
+| PID                                | 1096190                                 |
+|     Avg SM Utilization (%)         | 14                                      |
+|     Avg Memory Utilization (%)     | 2                                       |
++-----  Overall Health  -------------+-----------------------------------------+
+| Overall Health                     | Healthy                                 |
++------------------------------------+-----------------------------------------+
+

demo_v2.py

@@ -0,0 +1,648 @@
+# python demo_v2.py --cfg-path eval_configs/minigptv2_eval.yaml --gpu-id 0
+
+import argparse
+import os
+import random
+from collections import defaultdict
+
+import cv2
+import re
+
+import numpy as np
+from PIL import Image
+import torch
+import html
+import gradio as gr
+
+import torchvision.transforms as T
+import torch.backends.cudnn as cudnn
+
+from minigpt4.common.config import Config
+
+from minigpt4.common.registry import registry
+from minigpt4.conversation.conversation import Conversation, SeparatorStyle, Chat
+
+# imports modules for registration
+from minigpt4.datasets.builders import *
+from minigpt4.models import *
+from minigpt4.processors import *
+from minigpt4.runners import *
+from minigpt4.tasks import *
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="Demo")
+    parser.add_argument("--cfg-path", default='eval_configs/minigptv2_eval.yaml',
+                        help="path to configuration file.")
+    parser.add_argument("--gpu-id", type=int, default=0, help="specify the gpu to load the model.")
+    parser.add_argument(
+        "--options",
+        nargs="+",
+        help="override some settings in the used config, the key-value pair "
+             "in xxx=yyy format will be merged into config file (deprecate), "
+             "change to --cfg-options instead.",
+    )
+    args = parser.parse_args()
+    return args
+
+
+random.seed(42)
+np.random.seed(42)
+torch.manual_seed(42)
+
+cudnn.benchmark = False
+cudnn.deterministic = True
+
+print('Initializing Chat')
+args = parse_args()
+cfg = Config(args)
+
+device = 'cuda:{}'.format(args.gpu_id)
+
+model_config = cfg.model_cfg
+model_config.device_8bit = args.gpu_id
+model_cls = registry.get_model_class(model_config.arch)
+model = model_cls.from_config(model_config).to(device)
+bounding_box_size = 100
+
+vis_processor_cfg = cfg.datasets_cfg.cc_sbu_align.vis_processor.train
+vis_processor = registry.get_processor_class(vis_processor_cfg.name).from_config(vis_processor_cfg)
+
+model = model.eval()
+
+CONV_VISION = Conversation(
+    system="",
+    roles=(r"<s>[INST] ", r" [/INST]"),
+    messages=[],
+    offset=2,
+    sep_style=SeparatorStyle.SINGLE,
+    sep="",
+)
+
+
+def extract_substrings(string):
+    # first check if there is no-finished bracket
+    index = string.rfind('}')
+    if index != -1:
+        string = string[:index + 1]
+
+    pattern = r'<p>(.*?)\}(?!<)'
+    matches = re.findall(pattern, string)
+    substrings = [match for match in matches]
+
+    return substrings
+
+
+def is_overlapping(rect1, rect2):
+    x1, y1, x2, y2 = rect1
+    x3, y3, x4, y4 = rect2
+    return not (x2 < x3 or x1 > x4 or y2 < y3 or y1 > y4)
+
+
+def computeIoU(bbox1, bbox2):
+    x1, y1, x2, y2 = bbox1
+    x3, y3, x4, y4 = bbox2
+    intersection_x1 = max(x1, x3)
+    intersection_y1 = max(y1, y3)
+    intersection_x2 = min(x2, x4)
+    intersection_y2 = min(y2, y4)
+    intersection_area = max(0, intersection_x2 - intersection_x1 + 1) * max(0, intersection_y2 - intersection_y1 + 1)
+    bbox1_area = (x2 - x1 + 1) * (y2 - y1 + 1)
+    bbox2_area = (x4 - x3 + 1) * (y4 - y3 + 1)
+    union_area = bbox1_area + bbox2_area - intersection_area
+    iou = intersection_area / union_area
+    return iou
+
+
+def save_tmp_img(visual_img):
+    file_name = "".join([str(random.randint(0, 9)) for _ in range(5)]) + ".jpg"
+    file_path = "/tmp/gradio" + file_name
+    visual_img.save(file_path)
+    return file_path
+
+
+def mask2bbox(mask):
+    if mask is None:
+        return ''
+    mask = mask.resize([100, 100], resample=Image.NEAREST)
+    mask = np.array(mask)[:, :, 0]
+
+    rows = np.any(mask, axis=1)
+    cols = np.any(mask, axis=0)
+
+    if rows.sum():
+        # Get the top, bottom, left, and right boundaries
+        rmin, rmax = np.where(rows)[0][[0, -1]]
+        cmin, cmax = np.where(cols)[0][[0, -1]]
+        bbox = '{{<{}><{}><{}><{}>}}'.format(cmin, rmin, cmax, rmax)
+    else:
+        bbox = ''
+
+    return bbox
+
+
+def escape_markdown(text):
+    # List of Markdown special characters that need to be escaped
+    md_chars = ['<', '>']
+
+    # Escape each special character
+    for char in md_chars:
+        text = text.replace(char, '\\' + char)
+
+    return text
+
+
+def reverse_escape(text):
+    md_chars = ['\\<', '\\>']
+
+    for char in md_chars:
+        text = text.replace(char, char[1:])
+
+    return text
+
+
+colors = [
+    (255, 0, 0),
+    (0, 255, 0),
+    (0, 0, 255),
+    (210, 210, 0),
+    (255, 0, 255),
+    (0, 255, 255),
+    (114, 128, 250),
+    (0, 165, 255),
+    (0, 128, 0),
+    (144, 238, 144),
+    (238, 238, 175),
+    (255, 191, 0),
+    (0, 128, 0),
+    (226, 43, 138),
+    (255, 0, 255),
+    (0, 215, 255),
+]
+
+color_map = {
+    f"{color_id}": f"#{hex(color[2])[2:].zfill(2)}{hex(color[1])[2:].zfill(2)}{hex(color[0])[2:].zfill(2)}" for
+    color_id, color in enumerate(colors)
+}
+
+used_colors = colors
+
+
+def visualize_all_bbox_together(image, generation):
+    if image is None:
+        return None, ''
+
+    generation = html.unescape(generation)
+    print('gen begin', generation)
+    image_width, image_height = image.size
+    image = image.resize([500, int(500 / image_width * image_height)])
+    image_width, image_height = image.size
+
+    string_list = extract_substrings(generation)
+    if string_list:  # it is grounding or detection
+        mode = 'all'
+        entities = defaultdict(list)
+        i = 0
+        j = 0
+        for string in string_list:
+            try:
+                obj, string = string.split('</p>')
+            except ValueError:
+                print('wrong string: ', string)
+                continue
+            bbox_list = string.split('<delim>')
+            flag = False
+            for bbox_string in bbox_list:
+                integers = re.findall(r'-?\d+', bbox_string)
+                if len(integers) == 4:
+                    x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
+                    left = x0 / bounding_box_size * image_width
+                    bottom = y0 / bounding_box_size * image_height
+                    right = x1 / bounding_box_size * image_width
+                    top = y1 / bounding_box_size * image_height
+
+                    entities[obj].append([left, bottom, right, top])
+
+                    j += 1
+                    flag = True
+            if flag:
+                i += 1
+    else:
+        integers = re.findall(r'-?\d+', generation)
+
+        if len(integers) == 4:  # it is refer
+            mode = 'single'
+
+            entities = list()
+            x0, y0, x1, y1 = int(integers[0]), int(integers[1]), int(integers[2]), int(integers[3])
+            left = x0 / bounding_box_size * image_width
+            bottom = y0 / bounding_box_size * image_height
+            right = x1 / bounding_box_size * image_width
+            top = y1 / bounding_box_size * image_height
+            entities.append([left, bottom, right, top])
+        else:
+            # don't detect any valid bbox to visualize
+            return None, ''
+
+    if len(entities) == 0:
+        return None, ''
+
+    if isinstance(image, Image.Image):
+        image_h = image.height
+        image_w = image.width
+        image = np.array(image)
+
+    elif isinstance(image, str):
+        if os.path.exists(image):
+            pil_img = Image.open(image).convert("RGB")
+            image = np.array(pil_img)[:, :, [2, 1, 0]]
+            image_h = pil_img.height
+            image_w = pil_img.width
+        else:
+            raise ValueError(f"invaild image path, {image}")
+    elif isinstance(image, torch.Tensor):
+
+        image_tensor = image.cpu()
+        reverse_norm_mean = torch.tensor([0.48145466, 0.4578275, 0.40821073])[:, None, None]
+        reverse_norm_std = torch.tensor([0.26862954, 0.26130258, 0.27577711])[:, None, None]
+        image_tensor = image_tensor * reverse_norm_std + reverse_norm_mean
+        pil_img = T.ToPILImage()(image_tensor)
+        image_h = pil_img.height
+        image_w = pil_img.width
+        image = np.array(pil_img)[:, :, [2, 1, 0]]
+    else:
+        raise ValueError(f"invaild image format, {type(image)} for {image}")
+
+    indices = list(range(len(entities)))
+
+    new_image = image.copy()
+
+    previous_bboxes = []
+    # size of text
+    text_size = 0.5
+    # thickness of text
+    text_line = 1  # int(max(1 * min(image_h, image_w) / 512, 1))
+    box_line = 2
+    (c_width, text_height), _ = cv2.getTextSize("F", cv2.FONT_HERSHEY_COMPLEX, text_size, text_line)
+    base_height = int(text_height * 0.675)
+    text_offset_original = text_height - base_height
+    text_spaces = 2
+
+    # num_bboxes = sum(len(x[-1]) for x in entities)
+    used_colors = colors  # random.sample(colors, k=num_bboxes)
+
+    color_id = -1
+    for entity_idx, entity_name in enumerate(entities):
+        if mode == 'single' or mode == 'identify':
+            bboxes = entity_name
+            bboxes = [bboxes]
+        else:
+            bboxes = entities[entity_name]
+        color_id += 1
+        for bbox_id, (x1_norm, y1_norm, x2_norm, y2_norm) in enumerate(bboxes):
+            skip_flag = False
+            orig_x1, orig_y1, orig_x2, orig_y2 = int(x1_norm), int(y1_norm), int(x2_norm), int(y2_norm)
+
+            color = used_colors[entity_idx % len(used_colors)]  # tuple(np.random.randint(0, 255, size=3).tolist())
+            new_image = cv2.rectangle(new_image, (orig_x1, orig_y1), (orig_x2, orig_y2), color, box_line)
+
+            if mode == 'all':
+                l_o, r_o = box_line // 2 + box_line % 2, box_line // 2 + box_line % 2 + 1
+
+                x1 = orig_x1 - l_o
+                y1 = orig_y1 - l_o
+
+                if y1 < text_height + text_offset_original + 2 * text_spaces:
+                    y1 = orig_y1 + r_o + text_height + text_offset_original + 2 * text_spaces
+                    x1 = orig_x1 + r_o
+
+                # add text background
+                (text_width, text_height), _ = cv2.getTextSize(f"  {entity_name}", cv2.FONT_HERSHEY_COMPLEX, text_size,
+                                                               text_line)
+                text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2 = x1, y1 - (
+                            text_height + text_offset_original + 2 * text_spaces), x1 + text_width, y1
+
+                for prev_bbox in previous_bboxes:
+                    if computeIoU((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']) > 0.95 and \
+                            prev_bbox['phrase'] == entity_name:
+                        skip_flag = True
+                        break
+                    while is_overlapping((text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), prev_bbox['bbox']):
+                        text_bg_y1 += (text_height + text_offset_original + 2 * text_spaces)
+                        text_bg_y2 += (text_height + text_offset_original + 2 * text_spaces)
+                        y1 += (text_height + text_offset_original + 2 * text_spaces)
+
+                        if text_bg_y2 >= image_h:
+                            text_bg_y1 = max(0, image_h - (text_height + text_offset_original + 2 * text_spaces))
+                            text_bg_y2 = image_h
+                            y1 = image_h
+                            break
+                if not skip_flag:
+                    alpha = 0.5
+                    for i in range(text_bg_y1, text_bg_y2):
+                        for j in range(text_bg_x1, text_bg_x2):
+                            if i < image_h and j < image_w:
+                                if j < text_bg_x1 + 1.35 * c_width:
+                                    # original color
+                                    bg_color = color
+                                else:
+                                    # white
+                                    bg_color = [255, 255, 255]
+                                new_image[i, j] = (alpha * new_image[i, j] + (1 - alpha) * np.array(bg_color)).astype(
+                                    np.uint8)
+
+                    cv2.putText(
+                        new_image, f"  {entity_name}", (x1, y1 - text_offset_original - 1 * text_spaces),
+                        cv2.FONT_HERSHEY_COMPLEX, text_size, (0, 0, 0), text_line, cv2.LINE_AA
+                    )
+
+                    previous_bboxes.append(
+                        {'bbox': (text_bg_x1, text_bg_y1, text_bg_x2, text_bg_y2), 'phrase': entity_name})
+
+    if mode == 'all':
+        def color_iterator(colors):
+            while True:
+                for color in colors:
+                    yield color
+
+        color_gen = color_iterator(colors)
+
+        # Add colors to phrases and remove <p></p>
+        def colored_phrases(match):
+            phrase = match.group(1)
+            color = next(color_gen)
+            return f'<span style="color:rgb{color}">{phrase}</span>'
+
+        generation = re.sub(r'{<\d+><\d+><\d+><\d+>}|<delim>', '', generation)
+        generation_colored = re.sub(r'<p>(.*?)</p>', colored_phrases, generation)
+    else:
+        generation_colored = ''
+
+    pil_image = Image.fromarray(new_image)
+    return pil_image, generation_colored
+
+
+def gradio_reset(chat_state, img_list):
+    if chat_state is not None:
+        chat_state.messages = []
+    if img_list is not None:
+        img_list = []
+    return None, gr.update(value=None, interactive=True), gr.update(placeholder='Upload your image and chat',
+                                                                    interactive=True), chat_state, img_list
+
+
+def image_upload_trigger(upload_flag, replace_flag, img_list):
+    # set the upload flag to true when receive a new image.
+    # if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
+    upload_flag = 1
+    if img_list:
+        replace_flag = 1
+    return upload_flag, replace_flag
+
+
+def example_trigger(text_input, image, upload_flag, replace_flag, img_list):
+    # set the upload flag to true when receive a new image.
+    # if there is an old image (and old conversation), set the replace flag to true to reset the conv later.
+    upload_flag = 1
+    if img_list or replace_flag == 1:
+        replace_flag = 1
+
+    return upload_flag, replace_flag
+
+
+def gradio_ask(user_message, chatbot, chat_state, gr_img, img_list, upload_flag, replace_flag):
+    if len(user_message) == 0:
+        text_box_show = 'Input should not be empty!'
+    else:
+        text_box_show = ''
+
+    if isinstance(gr_img, dict):
+        gr_img, mask = gr_img['image'], gr_img['mask']
+    else:
+        mask = None
+
+    if '[identify]' in user_message:
+        # check if user provide bbox in the text input
+        integers = re.findall(r'-?\d+', user_message)
+        if len(integers) != 4:  # no bbox in text
+            bbox = mask2bbox(mask)
+            user_message = user_message + bbox
+
+    if chat_state is None:
+        chat_state = CONV_VISION.copy()
+
+    if upload_flag:
+        if replace_flag:
+            chat_state = CONV_VISION.copy()  # new image, reset everything
+            replace_flag = 0
+            chatbot = []
+        img_list = []
+        llm_message = chat.upload_img(gr_img, chat_state, img_list)
+        upload_flag = 0
+
+    chat.ask(user_message, chat_state)
+
+    chatbot = chatbot + [[user_message, None]]
+
+    if '[identify]' in user_message:
+        visual_img, _ = visualize_all_bbox_together(gr_img, user_message)
+        if visual_img is not None:
+            file_path = save_tmp_img(visual_img)
+            chatbot = chatbot + [[(file_path,), None]]
+
+    return text_box_show, chatbot, chat_state, img_list, upload_flag, replace_flag
+
+
+def gradio_answer(chatbot, chat_state, img_list, temperature):
+    llm_message = chat.answer(conv=chat_state,
+                              img_list=img_list,
+                              temperature=temperature,
+                              max_new_tokens=500,
+                              max_length=2000)[0]
+    chatbot[-1][1] = llm_message
+    return chatbot, chat_state
+
+
+def gradio_stream_answer(chatbot, chat_state, img_list, temperature):
+    if len(img_list) > 0:
+        if not isinstance(img_list[0], torch.Tensor):
+            chat.encode_img(img_list)
+    streamer = chat.stream_answer(conv=chat_state,
+                                  img_list=img_list,
+                                  temperature=temperature,
+                                  max_new_tokens=500,
+                                  max_length=2000)
+    output = ''
+    for new_output in streamer:
+        escapped = escape_markdown(new_output)
+        output += escapped
+        chatbot[-1][1] = output
+        yield chatbot, chat_state
+    chat_state.messages[-1][1] = '</s>'
+    return chatbot, chat_state
+
+
+def gradio_visualize(chatbot, gr_img):
+    if isinstance(gr_img, dict):
+        gr_img, mask = gr_img['image'], gr_img['mask']
+
+    unescaped = reverse_escape(chatbot[-1][1])
+    visual_img, generation_color = visualize_all_bbox_together(gr_img, unescaped)
+    if visual_img is not None:
+        if len(generation_color):
+            chatbot[-1][1] = generation_color
+        file_path = save_tmp_img(visual_img)
+        chatbot = chatbot + [[None, (file_path,)]]
+
+    return chatbot
+
+
+def gradio_taskselect(idx):
+    prompt_list = [
+        '',
+        '[grounding] describe this image in detail',
+        '[refer] ',
+        '[detection] ',
+        '[identify] what is this ',
+        '[vqa] '
+    ]
+    instruct_list = [
+        '**Hint:** Type in whatever you want',
+        '**Hint:** Send the command to generate a grounded image description',
+        '**Hint:** Type in a phrase about an object in the image and send the command',
+        '**Hint:** Type in a caption or phrase, and see object locations in the image',
+        '**Hint:** Draw a bounding box on the uploaded image then send the command. Click the "clear" botton on the top right of the image before redraw',
+        '**Hint:** Send a question to get a short answer',
+    ]
+    return prompt_list[idx], instruct_list[idx]
+
+
+
+
+chat = Chat(model, vis_processor, device=device)
+
+title = """<h1 align="center">MiniGPT-Med Demo</h1>"""
+description = 'Welcome to Our MiniGPT-Med Chatbot Demo!'
+# article = """<p><a href='https://minigpt-v2.github.io'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p><p><a href='https://github.com/Vision-CAIR/MiniGPT-4/blob/main/MiniGPTv2.pdf'><img src='https://img.shields.io/badge/Paper-PDF-red'></a></p><p><a href='https://github.com/Vision-CAIR/MiniGPT-4'><img src='https://img.shields.io/badge/GitHub-Repo-blue'></a></p><p><a href='https://www.youtube.com/watch?v=atFCwV2hSY4'><img src='https://img.shields.io/badge/YouTube-Video-red'></a></p>"""
+article = """<p><a href='https://minigpt-med.github.io/'><img src='https://img.shields.io/badge/Project-Page-Green'></a></p>"""
+
+introduction = '''
+For Abilities Involving Visual Grounding:
+1. Grounding: CLICK **Send** to generate a grounded image description.
+2. Refer: Input a referring object and CLICK **Send**.
+3. Detection: Write a caption or phrase, and CLICK **Send**.
+4. Identify: Draw the bounding box on the uploaded image window and CLICK **Send** to generate the bounding box. (CLICK "clear" button before re-drawing next time).
+5. VQA: Input a visual question and CLICK **Send**.
+6. No Tag: Input whatever you want and CLICK **Send** without any tagging
+
+You can also simply chat in free form!
+'''
+
+text_input = gr.Textbox(placeholder='Upload your image and chat', interactive=True, show_label=False, container=False,
+                        scale=8)
+with gr.Blocks() as demo:
+    gr.Markdown(title)
+    # gr.Markdown(description)
+    gr.Markdown(article)
+
+    with gr.Row():
+        with gr.Column(scale=0.5):
+            image = gr.Image(type="pil", tool='sketch', brush_radius=20)
+
+            temperature = gr.Slider(
+                minimum=0.1,
+                maximum=1.5,
+                value=0.6,
+                step=0.1,
+                interactive=True,
+                label="Temperature",
+            )
+
+            clear = gr.Button("Restart")
+
+            gr.Markdown(introduction)
+
+        with gr.Column():
+            chat_state = gr.State(value=None)
+            img_list = gr.State(value=[])
+            chatbot = gr.Chatbot(label='MiniGPT-Med')
+
+            dataset = gr.Dataset(
+                components=[gr.Textbox(visible=False)],
+                samples=[['No Tag'], ['Grounding'], ['Refer'], ['Detection'], ['Identify'], ['VQA']],
+                type="index",
+                label='Task Shortcuts',
+            )
+            task_inst = gr.Markdown('**Hint:** Upload your image and chat')
+            with gr.Row():
+                text_input.render()
+                send = gr.Button("Send", variant='primary', size='sm', scale=1)
+
+    upload_flag = gr.State(value=0)
+    replace_flag = gr.State(value=0)
+    image.upload(image_upload_trigger, [upload_flag, replace_flag, img_list], [upload_flag, replace_flag])
+# [29, 44, 42, 56]
+    with gr.Row():
+        with gr.Column():
+            gr.Examples(examples=[
+                ["Med_examples_v2/xmlab149/source.jpg", "[identify] what is this {<56><16><84><58>}", upload_flag,
+                 replace_flag, img_list],
+                ["Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.1495.1517874291.249176.jpg", "[detection] pneumonia", upload_flag, replace_flag, img_list],
+                ["Med_examples_v2/1.2.840.113654.2.55.48339325922382839066544590341580673064.png", "[refer] the nodule in the left lung", upload_flag, replace_flag,
+                 img_list],
+               ["Med_examples_v2/xmlab589/source.jpg", "[grounding] describe this image in detail", upload_flag, replace_flag, img_list],
+            ], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
+                outputs=[upload_flag, replace_flag])
+        with gr.Column():
+            gr.Examples(examples=[
+                ["Med_examples_v2/synpic50958.jpg", "[vqa] What does the small white lesions in the aorta mean?",
+                 upload_flag, replace_flag, img_list],
+                ["Med_examples_v2/5f4e8079-8225a5d2-1b0c3c46-4394a094-f285db0e.jpg", "Please provide a detailed description of the picture", upload_flag, replace_flag, img_list],
+                ["Med_examples_v2/1.2.276.0.7230010.3.1.4.8323329.16254.1517874395.786150.jpg", "Diagnose this image", upload_flag, replace_flag, img_list],
+                ["Med_examples_v2/synpic58547.jpg", "Could you describe the contents of this image for me?", upload_flag,
+                 replace_flag, img_list],
+            ], inputs=[image, text_input, upload_flag, replace_flag, img_list], fn=example_trigger,
+                outputs=[upload_flag, replace_flag])
+
+    dataset.click(
+        gradio_taskselect,
+        inputs=[dataset],
+        outputs=[text_input, task_inst],
+        show_progress="hidden",
+        postprocess=False,
+        queue=False,
+    )
+
+    text_input.submit(
+        gradio_ask,
+        [text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
+        [text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
+    ).success(
+        gradio_stream_answer,
+        [chatbot, chat_state, img_list, temperature],
+        [chatbot, chat_state]
+    ).success(
+        gradio_visualize,
+        [chatbot, image],
+        [chatbot],
+        queue=False,
+    )
+
+    send.click(
+        gradio_ask,
+        [text_input, chatbot, chat_state, image, img_list, upload_flag, replace_flag],
+        [text_input, chatbot, chat_state, img_list, upload_flag, replace_flag], queue=False
+    ).success(
+        gradio_stream_answer,
+        [chatbot, chat_state, img_list, temperature],
+        [chatbot, chat_state]
+    ).success(
+        gradio_visualize,
+        [chatbot, image],
+        [chatbot],
+        queue=False,
+    )
+
+    clear.click(gradio_reset, [chat_state, img_list], [chatbot, image, text_input, chat_state, img_list], queue=False)
+
+demo.launch(share=True, enable_queue=True)

environment.yml

@@ -0,0 +1,35 @@
+name: miniGPT-Med
+channels:
+  - pytorch
+  - defaults
+  - anaconda
+dependencies:
+  - python=3.9
+  - cudatoolkit
+  - pip
+  - pip:
+    - torch==2.0.0
+    - torchaudio
+    - torchvision
+    - huggingface-hub==0.18.0
+    - matplotlib==3.7.0
+    - psutil==5.9.4
+    - iopath
+    - pyyaml==6.0
+    - regex==2022.10.31
+    - tokenizers==0.13.2
+    - tqdm==4.64.1
+    - transformers==4.30.0
+    - timm==0.6.13
+    - webdataset==0.2.48
+    - omegaconf==2.3.0
+    - opencv-python==4.7.0.72
+    - decord==0.6.0
+    - peft==0.2.0
+    - sentence-transformers
+    - gradio==3.47.1
+    - accelerate==0.20.3
+    - bitsandbytes==0.37.0
+    - scikit-image
+    - visual-genome
+    - wandb

eval_configs/minigptv2_benchmark_evaluation.yaml

@@ -0,0 +1,69 @@
+model:
+  arch: minigpt_v2
+  model_type: pretrain
+  max_txt_len: 500
+  end_sym: "</s>"
+  low_resource: False
+  prompt_template: '[INST] {} [/INST]'
+  llama_model: "/ibex/project/c2106/RadGPT/MiniGPT4-v2/llama-2-7b-chat-hf"
+  ckpt: "/ibex/project/c2106/RadGPT/MiniGPT-Med-github/miniGPT_Med_.pth"
+  lora_r: 64
+  lora_alpha: 16
+
+datasets:
+  cc_sbu_align:
+    vis_processor:
+      train:
+        name: "blip2_image_eval"
+        image_size: 448
+    text_processor:
+      train:
+        name: "blip_caption"
+
+evaluation_datasets:
+  rsna:
+    eval_file_path: miniGPT-Med/json_files/RSNA/full_RSNA_1024.json
+    img_path: miniGPT-Med/RSNA/RSNA-bbox-1024
+    max_new_tokens: 100
+    batch_size: 10
+
+  radvqa:
+    eval_file_path: /miniGPT-Med/json_files/vqa/full_radVQA.json
+    img_path: /miniGPT-Med/radVQA/VQA_RAD_Images
+    max_new_tokens: 300
+    batch_size: 10
+
+  mimic_cxr:
+    eval_file_path: /miniGPT-Med/json_files/mimic/MIMIC_test.json
+    img_path: /miniGPT-Med/mimic-cxr-dataset/image
+    max_new_tokens: 300
+    batch_size: 10
+    
+  nlst:
+    eval_file_path: /miniGPT-Med/json_files/NLST/NLST_test.json
+    img_path: /miniGPT-Med/NLST/NLST_images
+    max_new_tokens: 100
+    batch_size: 10
+
+  detect_mimic:
+    eval_file_path: /miniGPT-Med/json_files/MIMIC-bbox/MIMIC-benchmarck.json
+    img_path: /miniGPT-Med/mimic-cxr-dataset/image
+    max_new_tokens: 100
+    batch_size: 10
+
+  SLAKE:
+    eval_file_path: /miniGPT-Med/json_files/SLAKE/grounding_test_SLAKE.json
+    img_path: /miniGPT-Med/SLAKE_images/imgs
+    max_new_tokens: 100
+    batch_size: 10
+
+
+run:
+  task: image_text_pretrain
+  name: minigptv2_evaluation
+  save_path: /miniGPT-Med/expermints
+
+  
+
+  
+

eval_configs/minigptv2_eval.yaml

@@ -0,0 +1,24 @@
+model:
+  arch: minigpt_v2
+  model_type: pretrain
+  max_txt_len: 500
+  end_sym: "</s>"
+  low_resource: True
+  prompt_template: '[INST] {} [/INST]'
+  ckpt: "/ibex/project/c2106/RadGPT/MiniGPT-Med-github/miniGPT_Med_.pth"
+  lora_r: 64
+  lora_alpha: 16
+
+
+datasets:
+  cc_sbu_align:
+    vis_processor:
+      train:
+        name: "blip2_image_eval"
+        image_size: 448
+    text_processor:
+      train:
+        name: "blip_caption"
+
+run:
+  task: image_text_pretrain

eval_scripts/clean_json.py

@@ -0,0 +1,74 @@
+import json
+import re 
+
+def clean_mimic_json(messy_json, cleaned_output):
+    with open(messy_json, 'r') as f:
+        messy_data = json.load(f)
+
+    clean_data = []
+    for image_id, captions in messy_data.items():
+        image_id_clean = image_id.split('.')[0]
+        caption_clean = ' '.join(captions)
+
+        clean_item = {
+            "image_id": image_id_clean,
+            "caption": caption_clean
+        }
+        
+        clean_data.append(clean_item)
+
+    with open(cleaned_output, 'w') as outfile:
+        json.dump(clean_data, outfile, indent=2)
+
+
+def clean_vqa_json(messy_json, cleaned_output):
+    with open(messy_json, "r") as file:
+        messy_json = json.load(file)
+
+    organized_json = {}
+
+    for key, values in messy_json.items():
+        organized_json[key] = []
+        for value in values:
+            organized_json[key].append({
+                "question": value["question"],
+                "answer": value["answer"]
+            })
+
+    with open(cleaned_output, "w") as outfile:
+        json.dump(organized_json, outfile, indent=4)
+
+
+
+def clean_detection_json(messy_json, cleaned_output):
+
+    with open(messy_json, "r") as input_file:
+        input_json = json.load(input_file)
+
+    organized_data = []
+
+    for key, value in input_json.items():
+        if value and isinstance(value, list) and len(value) > 0:
+            caption = value[0]
+            objects_match = caption.split("<p>")
+            if len(objects_match) == 2:
+                object_part = objects_match[1].split("</p>")[0].strip()
+            else:
+                object_part = ""
+            
+            bbox_match = re.findall(r'<(\d+)>', caption)
+            
+            if object_part and bbox_match and len(bbox_match) == 4:
+                key_part = key.split(".png")[0]
+                bbox_values = [float(val) for val in bbox_match]
+
+                organized_item = {
+                    "key": key_part,
+                    "objects": [object_part],
+                    "bbox": [bbox_values],
+                }
+
+                organized_data.append(organized_item)
+
+    with open(cleaned_output, "w") as output_file:
+        json.dump(organized_data, output_file, indent=4)

eval_scripts/metrics.py

@@ -0,0 +1,164 @@
+import sys
+sys.path.append('.')
+
+import json
+import pandas as pd
+import csv
+from sentence_transformers import SentenceTransformer, util
+from minigpt4.common.eval_utils import computeIoU
+
+# Load pre-trained BERT model
+model = SentenceTransformer('paraphrase-MiniLM-L6-v2')
+
+
+# BERT similarity function will be utilized in the two following functions
+def compute_bert_similarity(prediction_caption, ground_truth_caption):
+    prediction_embedding = model.encode([prediction_caption])
+    ground_truth_embedding = model.encode([ground_truth_caption])
+    similarity = util.pytorch_cos_sim(prediction_embedding, ground_truth_embedding)[0][0].item()
+    return similarity
+
+
+def MIMIC_BERT_Sim(gt_pth, pred_pth, output_csv):
+    # Read the ground truth and prediction JSON files
+    with open(gt_pth, 'r') as f:
+        ground_truth_data = json.load(f)
+    
+    with open(pred_pth, 'r') as f:
+        prediction_data = json.load(f)
+    
+    # Create a list to store BERT similarity data
+    bert_similarity_data = []
+    
+    # Initialize variables to calculate the average
+    total_similarity = 0
+    total_count = 0
+    
+    # Iterate over each item in the prediction_data list
+    for item in prediction_data:
+        # Extract the image_id and corresponding prediction caption
+        image_id = item["image_id"]
+        prediction_caption = item["caption"]
+        
+        # Search for the matching ground truth caption based on image_id
+        ground_truth_caption = None
+        for gt_item in ground_truth_data:
+            if gt_item["image_id"] == image_id:
+                ground_truth_caption = gt_item["caption"]
+                break
+        
+        if ground_truth_caption is not None:
+            bert_similarity = compute_bert_similarity(prediction_caption, ground_truth_caption)
+            bert_similarity_data.append({"image_id": image_id, "BERT_score": bert_similarity})
+            
+            total_similarity += bert_similarity
+            total_count += 1
+    
+    average_similarity = total_similarity / total_count if total_count > 0 else 0
+    
+    df = pd.DataFrame(bert_similarity_data)
+    df_sorted = df.sort_values(by="BERT_score", ascending=True)
+    df_sorted.to_csv(output_csv, index=False)
+    
+    return average_similarity
+
+def VQA_BERT_Sim(gt_pth, pred_pth, output_csv):
+    # Load ground truth JSON file
+    with open(gt_pth, 'r') as file:
+        gt_data = json.load(file)
+
+    # Load prediction JSON file
+    with open(pred_pth, 'r') as file:
+        prediction_data = json.load(file)
+
+    gt_qa_pairs = {(entry['image_name'], entry['question']): entry['answer'] for entry in gt_data}
+
+    def convert_to_dict(data):
+        qa_dict = {}
+        for image_name, qa_list in data.items():
+            for qa in qa_list:
+                key = (image_name, qa['question'])
+                qa_dict[key] = qa['answer']
+        return qa_dict
+
+    pred_qa_dict = convert_to_dict(prediction_data)
+
+    # Compute BERT similarity and create a list of results
+    results = []
+
+    for key, gt_answer in gt_qa_pairs.items():
+        if key in pred_qa_dict:
+            pred_answer = pred_qa_dict[key]
+            gt_answer = str(gt_answer)
+            pred_answer = str(pred_answer)
+
+            # Compute BERT similarity
+            similarity_score = compute_bert_similarity(pred_answer, gt_answer)
+
+            # Append the result to the list
+            results.append({
+                "img_name": key[0],
+                "question": key[1],
+                "answer": pred_answer,
+                "BERT_score": similarity_score
+            })
+
+    average_similarity = sum(entry["BERT_score"] for entry in results) / len(results) if results else 0
+    df = pd.DataFrame(results)
+    df_sorted = df.sort_values(by="BERT_score", ascending=True)
+    df_sorted.to_csv(output_csv, index=False)
+    print(f"Average BERT similarity score: {average_similarity}")
+
+
+#################################
+##############IoU################
+#################################
+
+def preprocess_bbox(bbox, original_size, image_size):
+    x1 = int((bbox[0] / original_size) * image_size)
+    y1 = int((bbox[1] / original_size) * image_size)
+    x2 = int((bbox[2] / original_size) * image_size)
+    y2 = int((bbox[3] / original_size) * image_size)
+    return [x1, y1, x2, y2]
+
+def average_iou(gt_pth, pred_pth, original_size, image_size, dataset_name, csv_filename):
+    # Load ground truth
+    with open(gt_pth, 'r') as file:
+        ground_truth = json.load(file)
+
+    # Load predictions
+    with open(pred_pth, 'r') as file:
+        predictions = json.load(file)
+
+    iou_list = []
+
+    with open(csv_filename, 'w', newline='') as csvfile:
+        fieldnames = ['image_name', 'IoU']
+        writer = csv.DictWriter(csvfile, fieldnames=fieldnames)
+        writer.writeheader()
+
+        for gt_item in ground_truth:
+            gt_key = gt_item['key']
+            gt_bboxes = gt_item['bbox']
+            original_size = gt_item['height']
+            gt_processed_bboxes = [preprocess_bbox(bbox, original_size, image_size) for bbox in gt_bboxes]
+
+            for pred_item in predictions:
+                pred_key = pred_item['key'].replace(".png", "")
+
+                if gt_key == pred_key:
+                    pred_bboxes = pred_item['bbox']
+                    try:
+                        for gt_bbox in gt_processed_bboxes:
+                            for pred_bbox in pred_bboxes:
+                                iou = computeIoU(gt_bbox, pred_bbox)
+                                iou_list.append(iou)
+                                writer.writerow({'image_name': gt_key, 'IoU': iou})
+                                print(gt_key)
+                                print(iou)
+                    except Exception as e:
+                        print("gt_bbox: ", gt_bbox)
+                        print("gt_bbox: ", pred_bboxes)
+
+    # average_iou = sum(iou_list) / len(iou_list)
+    # print(f"Average IoU for dataset {dataset_name}: {average_iou:.4f}")

eval_scripts/model_evaluation.py

@@ -0,0 +1,274 @@
+'''
+use this command in terminal to run the evaluation script
+torchrun --master-port 8888 --nproc_per_node 1 eval_scripts/model_evaluation.py  --cfg-path eval_configs/minigptv2_benchmark_evaluation.yaml --dataset 
+
+
+'''
+
+import sys
+sys.path.append('.')
+import os
+import re
+import json
+import argparse
+from collections import defaultdict
+import random
+import numpy as np
+from PIL import Image
+from tqdm import tqdm
+import torch
+from torch.utils.data import DataLoader
+from minigpt4.common.config import Config
+from minigpt4.common.eval_utils import prepare_texts, init_model, eval_parser, computeIoU
+from minigpt4.conversation.conversation import CONV_VISION_minigptv2
+
+from minigpt4.datasets.datasets.mimic_cxr_dataset import evalMIMICDataset, evalDetectMimicDataset
+from minigpt4.datasets.datasets.radvqa_dataset import evalRadVQADataset
+from minigpt4.datasets.datasets.nlst_dataset import eval_NLST_Dataset
+from minigpt4.datasets.datasets.rsna_dataset import evalRSNADataset
+from minigpt4.datasets.datasets.SLAKE_dataset import evalSLAKEDataset
+#import cleaning classes
+from eval_scripts.clean_json import clean_mimic_json, clean_vqa_json, clean_detection_json
+from eval_scripts.metrics import MIMIC_BERT_Sim, VQA_BERT_Sim, average_iou
+
+def list_of_str(arg):
+    return list(map(str, arg.split(',')))
+
+parser = eval_parser()
+parser.add_argument("--dataset", type=list_of_str, help="dataset to evaluate")
+
+args = parser.parse_args()
+
+cfg = Config(args)
+
+
+model, vis_processor = init_model(args)
+model.eval()
+CONV_VISION = CONV_VISION_minigptv2
+conv_temp = CONV_VISION.copy()
+conv_temp.system = ""
+model.eval()
+save_path = cfg.run_cfg.save_path
+
+def process_mimic_dataset():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+
+    with open((eval_file_path), 'r') as f:
+        mimic = json.load(f)
+            
+    data = evalMIMICDataset(mimic, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+        texts = prepare_texts(questions, conv_temp)  # warp the texts with conversation template
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+        for answer, img_id, question in zip(answers, img_ids, questions):
+            minigpt4_predict[img_id].append(answer)
+    
+    file_save_path = os.path.join(save_path,"MIMIC_inference_results_stage3.json")
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+    clean_mimic_json(file_save_path, file_save_path)
+
+    # csv file path to save the BERT results per each case
+    output_csv_path = '/miniGPT-Med/metric_results/bert_similarity_scores.csv'
+
+    # in MIMIC_BERT_Sim add the path of the ground_truth then the path of the inference result
+    average_similarity = MIMIC_BERT_Sim(eval_file_path, file_save_path, output_csv_path)
+    #print the average BERT_Sim
+    print("Average BERT Similarity:", average_similarity)
+
+def process_vqa_dataset():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+
+    with open((eval_file_path), 'r') as f:
+        radVQA = json.load(f)
+            
+    data = evalRadVQADataset(radVQA, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+        texts = prepare_texts(questions, conv_temp)  # warp the texts with conversation template
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+        for answer, img_id, question in zip(answers, img_ids, questions):
+            minigpt4_predict[img_id].append({"key":img_ids,"question": question.replace("[vqa]", "").strip() , "answer": answer})
+    
+    file_save_path = os.path.join(save_path,"radVQA_inference_results.json")
+    output_csv_path = '/miniGPT-Med/BERT_Sim_results/vqa_bert_similarity_scores.csv'
+
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+
+    clean_vqa_json(file_save_path, file_save_path)
+    VQA_BERT_Sim(eval_file_path, file_save_path, output_csv_path)
+
+def process_nlst_dataset():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"] 
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+
+    with open((eval_file_path), 'r') as f:
+        nlst = json.load(f)
+
+    data = eval_NLST_Dataset(nlst, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+    resamples = []
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+
+        texts = prepare_texts(questions, conv_temp)
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+
+        for answer, img_id, question in zip(answers, img_ids, questions):
+
+            # answer = answer.replace("<unk>","").replace(" ","").strip()
+            pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
+            minigpt4_predict[img_id].append(answer)
+
+    file_save_path = os.path.join(save_path,"NLST_inference_result.json")
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+    
+    csv_pth = os.path.join(save_path,"NLST_IoU_results.csv")
+    clean_detection_json(file_save_path,file_save_path)
+    average_iou(eval_file_path, file_save_path, 512, 100, "NLST", csv_pth)
+
+
+
+def process_rsna_dataset():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    print(eval_file_path)
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"] 
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+    print("----config----")
+    with open((eval_file_path), 'r') as f:
+        nlst = json.load(f)
+
+    data = evalRSNADataset(nlst, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+    resamples = []
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+        texts = prepare_texts(questions, conv_temp)
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+
+        for answer, img_id, question in zip(answers, img_ids, questions):
+
+            # answer = answer.replace("<unk>","").replace(" ","").strip()
+            pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
+            minigpt4_predict[img_id].append(answer)
+            print(img_id)
+            print(answer)
+
+    file_save_path = os.path.join(save_path,"RSNA_inference_result.json")
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+    
+    csv_pth = os.path.join(save_path,"RSNA_IoU_results.csv")
+    clean_detection_json(file_save_path,file_save_path)
+    average_iou(eval_file_path, file_save_path, 1024, 100, "rsna", csv_pth)
+
+
+def process_detect_mimic():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"] 
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+
+    with open((eval_file_path), 'r') as f:
+        nlst = json.load(f)
+
+    data = evalDetectMimicDataset(nlst, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+    resamples = []
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+
+        texts = prepare_texts(questions, conv_temp)
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+
+        for answer, img_id, question in zip(answers, img_ids, questions):
+            pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
+            minigpt4_predict[img_id].append(answer)
+
+    file_save_path = os.path.join(save_path,"Detect_MIMIC_inference_result.json")
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+    
+
+    csv_pth = os.path.join(save_path,"MIMIC_IoU_results.csv")
+    clean_detection_json(file_save_path,file_save_path)
+    average_iou(eval_file_path, file_save_path, "to be specified soon", 100, "MIMIC", csv_pth)
+
+
+
+def process_SLAKE_dataset():
+    eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
+    img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"] 
+    batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
+    max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
+
+    with open((eval_file_path), 'r') as f:
+        SLAKE = json.load(f)
+
+    data = evalSLAKEDataset(SLAKE, vis_processor, img_path)
+    eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
+    minigpt4_predict = defaultdict(list)
+    resamples = []
+
+    for images, questions, img_ids in tqdm(eval_dataloader):
+
+        texts = prepare_texts(questions, conv_temp)
+        answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
+
+        for answer, img_id, question in zip(answers, img_ids, questions):
+
+            # answer = answer.replace("<unk>","").replace(" ","").strip()
+            pattern = r'\{<\d{1,2}><\d{1,2}><\d{1,2}><\d{1,2}>\}'
+            minigpt4_predict[img_id].append(answer)
+
+    file_save_path = os.path.join(save_path,"SLAKE_inference_result.json")
+    with open(file_save_path,'w') as f:
+        json.dump(minigpt4_predict, f)
+    
+    csv_pth = os.path.join(save_path,"SLAKE_IoU_results.csv")
+    clean_detection_json(file_save_path,file_save_path)
+    average_iou(eval_file_path, file_save_path, 100, 100, "SLAKE", csv_pth)
+    
+
+
+############################################################################
+for dataset in args.dataset:
+    if dataset == 'mimic_cxr':
+        process_mimic_dataset()
+
+    elif dataset == 'radvqa':
+        process_vqa_dataset()
+
+    elif dataset == 'nlst':
+        process_nlst_dataset()
+
+    elif dataset == 'rsna':
+        process_rsna_dataset()
+
+    elif dataset == 'detect_mimic':
+        process_detect_mimic()
+
+    elif dataset == 'SLAKE':
+        process_SLAKE_dataset()
+
+    else:
+        print(f"Dataset '{dataset}' is not supported.")

miniGPTV2.yml

@@ -0,0 +1,35 @@
+name: GPTv2
+channels:
+  - pytorch
+  - defaults
+  - anaconda
+dependencies:
+  - python=3.9
+  - cudatoolkit
+  - pip
+  - pip:
+    - torch==2.0.0
+    - torchaudio
+    - torchvision
+    - huggingface-hub==0.18.0
+    - matplotlib==3.7.0
+    - psutil==5.9.4
+    - iopath
+    - pyyaml==6.0
+    - regex==2022.10.31
+    - tokenizers==0.13.2
+    - tqdm==4.64.1
+    - transformers==4.30.0
+    - timm==0.6.13
+    - webdataset==0.2.48
+    - omegaconf==2.3.0
+    - opencv-python==4.7.0.72
+    - decord==0.6.0
+    - peft==0.2.0
+    - sentence-transformers
+    - gradio==3.47.1
+    - accelerate==0.20.3
+    - bitsandbytes==0.37.0
+    - scikit-image
+    - visual-genome
+    - wandb

miniGPT_Med_.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca2d7fc37dc5330cdae927c8a3ff649c5919c726eccb05cae921fb997028b08e
+size 679780138

minigpt4/__init__.py

@@ -0,0 +1,31 @@
+"""
+ Copyright (c) 2022, salesforce.com, inc.
+ All rights reserved.
+ SPDX-License-Identifier: BSD-3-Clause
+ For full license text, see the LICENSE_Lavis file in the repo root or https://opensource.org/licenses/BSD-3-Clause
+"""
+
+import os
+import sys
+
+from omegaconf import OmegaConf
+
+from minigpt4.common.registry import registry
+
+from minigpt4.datasets.builders import *
+from minigpt4.models import *
+from minigpt4.processors import *
+from minigpt4.tasks import *
+
+
+root_dir = os.path.dirname(os.path.abspath(__file__))
+default_cfg = OmegaConf.load(os.path.join(root_dir, "configs/default.yaml"))
+
+registry.register_path("library_root", root_dir)
+repo_root = os.path.join(root_dir, "..")
+registry.register_path("repo_root", repo_root)
+cache_root = os.path.join(repo_root, default_cfg.env.cache_root)
+registry.register_path("cache_root", cache_root)
+
+registry.register("MAX_INT", sys.maxsize)
+registry.register("SPLIT_NAMES", ["train", "val", "test"])