trl-env 0.1.3

No project description provided

TRL-ENV

ENV

TRL is a convenient library to train large language model (LLM) using reinforcement learning (RL). However, it is still too new, the interface is not well-developed yet. rollout_func is a low-level interface to write your own rollout for RL and environment_factory is a high-level interface to train your model with external environemnt, however, how it parse the model output for tool use is uncleared and not documented.

TRL-ENV addresses the middle-level with a very simple environment interface

type Action = str
type Delta = str
type Seed = str

class Env(Protocol):
    last_step_reward: float
    alive: bool
    def reset(self, seed: Seed) -> Delta: ...
    def step(self, action: Action) -> Delta: ...

It is similar to tool call if not the same. Note that, rollout_func is an experimental feature of TRL, this library is subject to break at anytime

It is important to note that, batch_rollout assumes the additivity of tokenizer, that is

tok(a ++ b) = tok(a) ++ tok(b)

where a and b are texts and ++ is concatenation. This is because Env interacts with LLM via text, not sequence of tokens.

PROCESSOR

transformers despite after 8 years of development (as of 2026) is still not stable. For example, not all models has Tokenizer.parse_response which should be a basic function that must be implemented from the beginning. TRL-ENV requires Tokenizer.parse_response to be existed by Processor interface

Language = str

class Processor(Protocol):
    def init_system_input(self, prompt: Language) -> str: ...
    def append_user_input(self, prompt: Language) -> str: ...
    def parse_agent_output(self, completion: Language) -> tuple[str, str]: ...

EXAMPLES

TRL-ENV provides a very simple example for training agentic LLM. See examples