refactor(Llama): enhance error handling and cleanup in eval method
#390
| Job | Run time |
|---|---|
| 2m 47s | |
| 2m 44s | |
| 6m 40s | |
| 7m 14s | |
| 6m 57s | |
| 3m 30s | |
| 3m 0s | |
| 2m 42s | |
| 4m 39s | |
| 3m 7s | |
| 2m 11s | |
| 6m 49s | |
| 5m 58s | |
| 2m 48s | |
| 2m 45s | |
| 2m 49s | |
| 1h 6m 40s |