AWARe on MLLM-DCL Benchmark
We trained this model on five tasks from the MLLM-DCL benchmark: RS, Med, AD, Sci, and Fin one by one. This checkpoint is the final one, obtained after training on Fin.
Below is the training loss curve:
The evaluation results are as follows:
| AWARe | RS | Med | AD | Sci | Fin |
|---|
| RS | 79.68 | | | | |
| Med | 80.12 | 59.23 | | | |
| AD | 80.38 | 59.23 | 53.57 | | |
| Sci | 79.31 | 54.71 | 52.34 | 52.53 | |
| Fin | 77.93 | 42.93 | 43.7 | 44.95 | 91.67 |
