Restore Agent's Last Exam with correct ale-cli-pass-rate task
#6
by SaylorTwift HF Staff - opened
.eval_results/Step-5-Preview.yaml
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
- dataset:
|
| 2 |
+
id: datacurve/deep-swe
|
| 3 |
+
task_id: deep_swe
|
| 4 |
+
value: 67.7
|
| 5 |
+
date: "2026-09-20"
|
| 6 |
+
source:
|
| 7 |
+
url: https://huggingface.co/stepfun-ai/Step-5-Preview
|
| 8 |
+
name: "Step 5 Preview model card"
|
| 9 |
+
notes: "DeepSWE 1.1, run with the SWE-agent harness."
|
| 10 |
+
- dataset:
|
| 11 |
+
id: harborframework/terminal-bench
|
| 12 |
+
task_id: terminalbench_4
|
| 13 |
+
value: 33.3
|
| 14 |
+
date: "2026-09-20"
|
| 15 |
+
source:
|
| 16 |
+
url: https://huggingface.co/stepfun-ai/Step-5-Preview
|
| 17 |
+
name: "Step 5 Preview model card"
|
| 18 |
+
- dataset:
|
| 19 |
+
id: agents-last-exam/agents-last-exam
|
| 20 |
+
task_id: ale-cli-pass-rate
|
| 21 |
+
value: 29.5
|
| 22 |
+
date: "2026-09-20"
|
| 23 |
+
source:
|
| 24 |
+
url: https://huggingface.co/stepfun-ai/Step-5-Preview
|
| 25 |
+
name: "Step 5 Preview model card"
|
| 26 |
+
notes: "ALE-CLI (CLI interface) split per the model card."
|