NousResearch/hermes-agent · error · ValueError

Unknown distribution: {distribution}. Available: {list(list_

Error message

Unknown distribution: {distribution}. Available: {list(list_distributions().keys())}

What it means

ValueError from the BatchRunner constructor (batch_runner.py:610): validate_distribution(distribution) rejected the requested sampling distribution, and the message enumerates the valid names via list_distributions(). It fails at construction time so an invalid run never starts.

Source

Thrown at batch_runner.py:610

        self.api_key = api_key
        self.model = model
        self.num_workers = num_workers
        self.verbose = verbose
        self.ephemeral_system_prompt = ephemeral_system_prompt
        self.log_prefix_chars = log_prefix_chars
        self.providers_allowed = providers_allowed
        self.providers_ignored = providers_ignored
        self.providers_order = providers_order
        self.provider_sort = provider_sort
        self.openrouter_min_coding_score = openrouter_min_coding_score
        self.max_tokens = max_tokens
        self.reasoning_config = reasoning_config
        self.prefill_messages = prefill_messages
        self.max_samples = max_samples
        
        # Validate distribution
        if not validate_distribution(distribution):
            raise ValueError(f"Unknown distribution: {distribution}. Available: {list(list_distributions().keys())}")
        
        # Setup output directory
        self.output_dir = Path("data") / run_name
        self.output_dir.mkdir(parents=True, exist_ok=True)
        
        # Checkpoint file
        self.checkpoint_file = self.output_dir / "checkpoint.json"
        
        # Statistics file
        self.stats_file = self.output_dir / "statistics.json"
        
        # Load dataset (and optionally truncate to max_samples)
        self.dataset = self._load_dataset()
        if self.max_samples and self.max_samples < len(self.dataset):
            full_count = len(self.dataset)
            self.dataset = self.dataset[:self.max_samples]
            print(f"✂️  Truncated dataset from {full_count} to {self.max_samples} samples (--max_samples)")
        

View on GitHub (pinned to c896c09c42)

Solutions

1. Call list_distributions() (its keys are printed in the message) and use one of those exact names.
2. Check for stray whitespace/case in the value coming from config or argv before constructing BatchRunner.
3. If you need a new distribution, register it with the distribution module so validate_distribution accepts it.

Example fix

# before
runner = BatchRunner(run_name="r1", distribution="unform", ...)

# after
from batch_runner import list_distributions  # or the distributions module
valid = list(list_distributions().keys())
assert distribution in valid, f"{distribution!r} not in {valid}"
runner = BatchRunner(run_name="r1", distribution="uniform", ...)

from batch_runner import list_distributions, validate_distribution

if not validate_distribution(distribution):
    raise SystemExit(
        f"unknown distribution {distribution!r}; choose one of {sorted(list_distributions())}"
    )

try:
    runner = BatchRunner(run_name=r, distribution=distribution, ...)
except ValueError as exc:
    raise SystemExit(str(exc)) from exc  # message already lists valid names

When it happens

Trigger: Instantiating BatchRunner with a distribution string that is misspelled, has wrong casing/whitespace, or is from an older/newer version's vocabulary — e.g. 'unform' instead of 'uniform'.

Common situations: CLI args or config files carrying a typo; scripts written against a different hermes version whose distribution names changed; copying an example that references a distribution that was renamed.

Related errors

• Invalid profile name: ${value}
• SSH host is required.
• Model {agent.model} has a context window of {_ctx:,} tokens,
• Malformed custom endpoint URL: {candidate!r}. Run `hermes se
• iron-proxy rejected the new config (validation failed; the r

AI-assisted analysis of NousResearch/hermes-agent@c896c09c42 (2026-08-14). Data as JSON: /api/errors/1d5fbfc3de2b9dfa. Report an issue: GitHub.