agentic-ptb
/

grok-record

Model card Files Files and versions

grok-record / harness /scripts /eval_both.sh

simonycl's picture

Upload folder using huggingface_hub

9b2f1cf verified 15 days ago

History Blame Contribute Delete

366 Bytes

	#!/usr/bin/env bash
	# Run both official suites against the currently served model (or $1).
	set -euo pipefail
	WS="${AGENTPTB_WORKSPACE:-/mnt/pvc/users/simon/agentptb/runs/grok/workspace}"
	MODEL="${1:-$WS/outputs/soup-85-s80/weights}"
	N="${2:-40}"
	"$WS/scripts/wait_api.sh"
	"$WS/scripts/launch_eval.sh" tb2 "$MODEL" "$N"
	"$WS/scripts/launch_eval.sh" swe "$MODEL" "$N"