### Install Stoa with All Adapters

Source: https://github.com/edantoledo/stoa/blob/main/README.md

Installs the 'stoa-env' library with all available environment adapters, providing support for all integrated RL libraries.

```Bash
pip install "stoa-env[all]"
```

--------------------------------

### Install Stoa Core Library

Source: https://github.com/edantoledo/stoa/blob/main/README.md

Installs the core 'stoa-env' library, which includes the main API and wrappers but excludes specific environment adapters.

```Bash
pip install stoa-env
```

--------------------------------

### Install Stoa with Brax Adapter

Source: https://github.com/edantoledo/stoa/blob/main/README.md

Installs the 'stoa-env' library along with the Brax environment adapter, allowing integration with Brax environments.

```Bash
pip install "stoa-env[brax]"
```

--------------------------------

### Install Stoa with Gymnax Adapter

Source: https://github.com/edantoledo/stoa/blob/main/README.md

Installs the 'stoa-env' library along with the Gymnax environment adapter, enabling the use of Gymnax environments within Stoa.

```Bash
pip install "stoa-env[gymnax]"
```

--------------------------------

### Adapt Gymnax Environment and Apply Wrappers in JAX

Source: https://github.com/edantoledo/stoa/blob/main/README.md

Demonstrates adapting a Gymnax environment to the Stoa interface, applying wrappers like AutoResetWrapper and RecordEpisodeMetrics, and JIT compiling reset and step functions for performance. It also shows basic interaction with the environment, including sampling actions and accessing episode metrics.

```Python
import jax
import gymnax
from stoa import GymnaxToStoa, FlattenObservationWrapper, AutoResetWrapper, RecordEpisodeMetrics

# 1. Instantiate a base environment from a supported library
gymnax_env, env_params = gymnax.make("CartPole-v1")

# 2. Adapt the environment to the Stoa interface
env = GymnaxToStoa(gymnax_env, env_params)

# 3. Apply standard wrappers
# Note: The order of wrappers matters.
env = AutoResetWrapper(env, next_obs_in_extras=True)
env = RecordEpisodeMetrics(env)

# JIT compile the reset and step functions for performance
env.reset = jax.jit(env.reset)
env.step = jax.jit(env.step)

# 4. Interact with the environment
rng_key = jax.random.PRNGKey(0)
state, timestep = env.reset(rng_key)
total_reward = 0

for _ in range(100):
    action = env.action_space().sample(rng_key)
    state, timestep = env.step(state, action)
    total_reward += timestep.reward

    if timestep.last():
        # Access metrics recorded by the RecordEpisodeMetrics wrapper
        episode_return = timestep.extras['episode_metrics']['episode_return']
        print(f"Episode finished. Return: {episode_return}")

        # The state has been auto-reset, so we can continue the loop
        total_reward = 0

```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.