The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Dory 4B single-call tool-use eval
Model: nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16
20 single-call tool-use tasks with stub (non-executing) functions, 4-5 tools offered per task. A run passes iff the model emits exactly the expected function name and an argument dict equal to the expected one after lenient normalization (dates, times, units, languages, numbers, whitespace, case). Sampling: temperature 1.0, top_p 0.95 (card recommendation).
| loops | pass share (raw) | pass share (artifact-adjusted) | runs |
|---|---|---|---|
| 1 | 90.0% (54/60) | 95.0% (57/60) | 60 |
| 4 | 95.0% (57/60) | 100.0% (60/60) | 60 |
Scoring note
The stub tool schemas mark every parameter required. Task t01 ("What's the weather
in Lisbon right now?") expected {city: Lisbon} with the optional unit omitted — but the
schema forces the model to supply unit, so the expected answer was unsatisfiable by
construction. The model consistently (6/6 runs, both loop counts) answered
{city: Lisbon, unit: celsius}. The adjusted column counts those as passes.
The only behavior difference between loop counts is t13: at 1 loop the model argued
{from_account: "checking account", to_account: "savings account"} (expected
"checking"/"savings") in all 3 trials; at 4 loops it emitted the exact strings 3/3.
Per-task pass rate
| task | prompt | loops 1 | loops 4 |
|---|---|---|---|
| t01 | What's the weather in Lisbon right now? | 0/3 | 0/3 |
| t02 | What's the weather in Chicago? Give it to me in Fahrenheit. | 3/3 | 3/3 |
| t03 | Can you pull up the 5-day forecast for Tokyo? | 3/3 | 3/3 |
| t04 | Convert 250 US dollars to euros for me. | 3/3 | 3/3 |
| t05 | Put a meeting called 'Design review' on my calendar for November 10, 2026 at 14:30, lasting 60 minutes. | 3/3 | 3/3 |
| t06 | Send an email to sarah@acme.com with the subject 'Q3 report' and the body 'The report is attached.' | 3/3 | 3/3 |
| t07 | What's Apple's current stock price? | 3/3 | 3/3 |
| t08 | Add 'Buy oat milk' to my groceries list, due November 5, 2026. | 3/3 | 3/3 |
| t09 | Find a Thai restaurant in Berlin for 4 people. | 3/3 | 3/3 |
| t10 | Calculate 18 * 24 + 7 for me. | 3/3 | 3/3 |
| t11 | Give me biking directions from Denver Union Station to the Denver Art Museum. | 3/3 | 3/3 |
| t12 | Remind me to call the dentist at 9:00 AM on December 1, 2026. | 3/3 | 3/3 |
| t13 | Transfer 120 dollars from my checking account to my savings account. | 0/3 | 3/3 |
| t14 | Can you check the status of my order A-48291? | 3/3 | 3/3 |
| t15 | Translate 'good morning' into Japanese. | 3/3 | 3/3 |
| t16 | Convert 72 degrees Fahrenheit to Celsius. | 3/3 | 3/3 |
| t17 | Book a hotel in Rome for 2 guests, checking in on November 20, 2026 and checking out on November 23, 2026. | 3/3 | 3/3 |
| t18 | Find flights from SFO to JFK on December 15, 2026. | 3/3 | 3/3 |
| t19 | Create an invoice for Northwind Traders for 480 dollars, due November 30, 2026. | 3/3 | 3/3 |
| t20 | Reserve a table for 6 people at Osteria Bianchi on November 14, 2026 at 19:00. | 3/3 | 3/3 |
Failures
- loops=1 t01 trial 0: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- loops=1 t01 trial 1: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- loops=1 t01 trial 2: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- loops=1 t13 trial 0: wrong_args (pred: transfer_money {"from_account": "checking account", "to_account": "savings account", "amount": 120})
- loops=1 t13 trial 1: wrong_args (pred: transfer_money {"from_account": "checking account", "to_account": "savings account", "amount": 120})
- loops=1 t13 trial 2: wrong_args (pred: transfer_money {"from_account": "checking account", "to_account": "savings account", "amount": 120})
- loops=4 t01 trial 0: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- loops=4 t01 trial 1: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- loops=4 t01 trial 2: wrong_args (pred: get_weather {"city": "Lisbon", "unit": "celsius"}) — scoring artifact, see note
- Downloads last month
- 72