tqchen commented on PR #15700:
URL: https://github.com/apache/tvm/pull/15700#issuecomment-1711783558
To expand a bit, since I think the examples are great illustrating the
overall case, there are usually two category of parameter inputs to a function
- C0: weights, usually fixed for all applications during inference
- C1: config settings(temperature, scaling factor) that are changeable
during inference.
A typical function signature should ideally separate C0 and C1
```python
def f(input, weight_params, config_params):
pass
```
So we will be able to transform and setup weight parameter once, while
passing config_params during runtime from the app configurations.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
