Model techniques map
Techniquespost-trainingreinforcement learning algorithm

implementation detail · filed under post-training

Per-Token Tool-Error Step Penalty

Applying a small negative reward to the tokens of an assistant step whose tool invocation caused an execution error.

Also called Tool-error step penalty.

source
1
model
1
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

for every assistant step whose tool invocation led to the error in tool execution, we apply a small negative reward to exactly the tokens that constituted that step, rather than to the trajectory as a whole.

usedtraining objectivein LAGUNA M.1/XS.2Poolside

Filed alongside

Other methods under post-training :: reinforcement learning algorithm.