# Use different lr for different policies

**URL:** <https://discuss.ray.io/t/use-different-lr-for-different-policies/14144>\
**Category:** RLlib\
**Created:** [March 25, 2024, 1:54am UTC](https://discuss.ray.io/t/use-different-lr-for-different-policies/14144 "2024-03-25T01:54:28Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![call-me-anything-you](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/call-me-anything-you/32/5519_2.png) [@call-me-anything-you](https://discuss.ray.io/u/call-me-anything-you)\
**Post date:** [March 25, 2024, 1:54am UTC](https://discuss.ray.io/t/use-different-lr-for-different-policies/14144/1 "2024-03-25T01:54:28Z")

</div>

I’m using rllib under the MARL setting and need to set different lr for different policies. I’ve tried to set these in configs.multi\_agent() as follows:

```python
configs.multi_agent(
    policies = {
        "policy_0": (None, None, None, {"lr": 1e-4}), 
        "policy_1": (None, None, None, {"lr": 2e-4}), 
    }
)

```

However, it seems like none of these works. The policies still use the lr set in

```python
configs.training(lr=xxx)

```

How can I do this right?