arithmetic-v2

Evaluate short integer expressions with standard precedence and give the exact result.

Field	Value
Reference	nodus/arithmetic-v2@2.0.0
Image	nodus/env-arithmetic-v2:2.0.0
Publisher	Nodus
Category	Math
Readiness	Stable
Modes	Train, Evaluate
Reward	Binary
Held-out measures	TrainedTask
Splits	500 train, 128 test (disjoint by canonical identity)
Licenses	code Apache-2.0, data Apache-2.0
Source	https://github.com/nodus-compute/nodus-platform/tree/main/images/environments

Grading

Completions are graded by the platform, never by the trainer: the trainer submits {taskId, completion} batches and the verdicts come back as task events.

Grader	Kind	What it checks
exact-integer	Program	The last N equals the expression’s value

Sample task

One line of nodus-env tasks --split test --seed 0; tasks never carry answers.

{
  "metadata": {
    "terms": 3
  },
  "prompt": "Compute 41 - 38 * 33. Give only the final integer, like \u003canswer\u003e42\u003c/answer\u003e.",
  "taskId": "arithmetic-v2:test:0:0"
}

Examples

Each example is a TrainingJob template. A baseline is shown only where the example was measured by running it.

Example	Mode	Runtime	Model	Tasks	Baseline	Trained	Measured on
arithmetic-grpo	Train	nodus/grpo-lora	Qwen/Qwen3-0.6B @ c1899de	64	not measured	not measured	not measured

Run one with a server dry-run first:

$ nodus create trainingjob my-run --from-example nodus/arithmetic-v2:arithmetic-grpo --dry-run=server -o estimate

import nodus

job = nodus.recipes.TrainingJob.from_example("nodus/arithmetic-v2:arithmetic-grpo")
plan = job.preview()
run = plan.run(max_cost=5)
print(run.wait().summary)