support finetune on pretrained models

[HuangJiameng]

WIP, building unittests.

[HuangJiameng]

unittests for usual condition is done

[codecov[bot]]

Codecov Report

Patch coverage: 82.85% and project coverage change: -0.14 :warning:

Comparison is base (b816df3) 87.94% compared to head (393e820) 87.80%.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #152 +/- ## ========================================== - Coverage 87.94% 87.80% -0.14% ========================================== Files 79 79 Lines 3791 3846 +55 ========================================== + Hits 3334 3377 +43 - Misses 457 469 +12 ``` | [Impacted Files](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling) | Coverage Δ | | |---|---|---| | [...pgen2/exploration/task/conf\_sampling\_task\_group.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2V4cGxvcmF0aW9uL3Rhc2svY29uZl9zYW1wbGluZ190YXNrX2dyb3VwLnB5) | `100.00% <ø> (ø)` | | | [dpgen2/flow/dpgen\_loop.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2Zsb3cvZHBnZW5fbG9vcC5weQ==) | `87.76% <ø> (ø)` | | | [dpgen2/fp/vasp\_input.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2ZwL3Zhc3BfaW5wdXQucHk=) | `93.50% <ø> (ø)` | | | [dpgen2/superop/block.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL3N1cGVyb3AvYmxvY2sucHk=) | `94.20% <ø> (ø)` | | | [dpgen2/entrypoint/submit.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2VudHJ5cG9pbnQvc3VibWl0LnB5) | `68.92% <38.88%> (-2.40%)` | :arrow_down: | | [dpgen2/op/run\_dp\_train.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL29wL3J1bl9kcF90cmFpbi5weQ==) | `98.41% <88.88%> (-0.49%)` | :arrow_down: | | [dpgen2/entrypoint/args.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2VudHJ5cG9pbnQvYXJncy5weQ==) | `91.20% <100.00%> (+0.21%)` | :arrow_up: | | [dpgen2/exploration/task/lmp\_template\_task\_group.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2V4cGxvcmF0aW9uL3Rhc2svbG1wX3RlbXBsYXRlX3Rhc2tfZ3JvdXAucHk=) | `95.55% <100.00%> (ø)` | | | [dpgen2/exploration/task/npt\_task\_group.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL2V4cGxvcmF0aW9uL3Rhc2svbnB0X3Rhc2tfZ3JvdXAucHk=) | `95.55% <100.00%> (ø)` | | | [dpgen2/op/select\_confs.py](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling#diff-ZHBnZW4yL29wL3NlbGVjdF9jb25mcy5weQ==) | `100.00% <100.00%> (ø)` | | | ... and [6 more](https://app.codecov.io/gh/deepmodeling/dpgen2/pull/152?src=pr&el=tree-more&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=deepmodeling) | |

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.

[njzjz]

@zjgemi, Could you take a look at the type checker? It is related to dflow.