Datasets:
key stringlengths 22 38 | intent stringlengths 11 168 | template_id stringclasses 11
values | seed int64 1 5k | traces listlengths 0 0 | candidates listlengths 3 3 |
|---|---|---|---|---|---|
trainer/trainer_actuator_pd/1 | PD actuators: kp 30.0, kd 4.0, torque limit 25.0 | trainer_actuator_pd | 1 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"30.0\", damping=\"4.0\", effort_limit=\"25.0\")\ndone = a.ENO\n",
"fbd_sha": "82fe6e065c1ad666d6bda95deae5d01e8fd1c542e1aab87bdc41ad5ccaf2f2fe",
"provenanc... |
trainer/trainer_actuator_pd/2 | uniform PD gains 20.0/1.0 on all actuators, clamped at 25.0 N m | trainer_actuator_pd | 2 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"20.0\", damping=\"1.0\", effort_limit=\"25.0\")\ndone = a.ENO\n",
"fbd_sha": "2c1df8205b53a379824a7560710f1c58292e0b9f1aab987e35f431b16aef6876",
"provenanc... |
trainer/trainer_actuator_pd/3 | drive every joint with a PD actuator of stiffness 30.0, damping 4.0 and an effort limit of 50.0 N m | trainer_actuator_pd | 3 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"30.0\", damping=\"4.0\", effort_limit=\"50.0\")\ndone = a.ENO\n",
"fbd_sha": "bbc1336aecc2d9c628f48c08643afda6988bdc8e2186eed3d4a0e4735c0718d2",
"provenanc... |
trainer/trainer_actuator_pd/4 | PD actuators: kp 30.0, kd 2.0, torque limit 25.0 | trainer_actuator_pd | 4 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"30.0\", damping=\"2.0\", effort_limit=\"25.0\")\ndone = a.ENO\n",
"fbd_sha": "f565e5fb63ab13988e65f02b9fe6afa0270cef34887ecf3e034f7294db7289c4",
"provenanc... |
trainer/trainer_actuator_pd/5 | uniform PD gains 80.0/2.0 on all actuators, clamped at 88.0 N m | trainer_actuator_pd | 5 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"80.0\", damping=\"2.0\", effort_limit=\"88.0\")\ndone = a.ENO\n",
"fbd_sha": "4e3b98bf9d3465980262044808e9d559c1d578808e29156f7c367cfd096c1d04",
"provenanc... |
trainer/trainer_actuator_pd/6 | drive every joint with a PD actuator of stiffness 80.0, damping 1.0 and an effort limit of 139.0 N m | trainer_actuator_pd | 6 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"80.0\", damping=\"1.0\", effort_limit=\"139.0\")\ndone = a.ENO\n",
"fbd_sha": "bb97c6514f2d13013996058e6071b071f29863d6128ee119177ffeecabab9d3c",
"provenan... |
trainer/trainer_actuator_pd/7 | PD actuators: kp 40.0, kd 1.5, torque limit 139.0 | trainer_actuator_pd | 7 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"40.0\", damping=\"1.5\", effort_limit=\"139.0\")\ndone = a.ENO\n",
"fbd_sha": "1c81c37a1a0766b9d96f7b00abbd3c7e822fdaae8c007e192446350fe7254622",
"provenan... |
trainer/trainer_actuator_pd/8 | uniform PD gains 30.0/2.0 on all actuators, clamped at 139.0 N m | trainer_actuator_pd | 8 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"30.0\", damping=\"2.0\", effort_limit=\"139.0\")\ndone = a.ENO\n",
"fbd_sha": "57f07045074ee4529bd6f39aecee433f7be50beb1045f6508397a5e0cde49cc0",
"provenan... |
trainer/trainer_actuator_pd/9 | drive every joint with a PD actuator of stiffness 60.0, damping 4.0 and an effort limit of 88.0 N m | trainer_actuator_pd | 9 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program actuator_pd\nvar done : BOOL\na = CALL[Actuator.pd](TARGET=\"actuator\", stiffness=\"60.0\", damping=\"4.0\", effort_limit=\"88.0\")\ndone = a.ENO\n",
"fbd_sha": "7d891ae29a3ff153a07fa63824f13d153e4c5fb1ee228c04e88979170a26a15c",
"provenanc... |
trainer/trainer_tracking/11 | velocity tracking rewards: lin 2.08/1.0, ang 2.46/1.0 | trainer_tracking | 11 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"2.08\", std=\"1.0\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.46\", std=\"1.0\")\ndone = b.ENO\n",
"fbd_sha": ... |
trainer/trainer_tracking/12 | reward linear velocity tracking at weight 2.16 with std 0.7071 and angular at 2.8 with std 0.5 | trainer_tracking | 12 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"2.16\", std=\"0.7071\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.8\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha"... |
trainer/trainer_tracking/13 | track the twist command: linear weight 1.41 (std 0.5), angular weight 2.9 (std 0.5) | trainer_tracking | 13 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"1.41\", std=\"0.5\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.9\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha": "... |
trainer/trainer_tracking/14 | velocity tracking rewards: lin 0.87/0.5, ang 2.96/0.7071 | trainer_tracking | 14 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"0.87\", std=\"0.5\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.96\", std=\"0.7071\")\ndone = b.ENO\n",
"fbd_sha... |
trainer/trainer_tracking/15 | reward linear velocity tracking at weight 3.88 with std 0.25 and angular at 0.54 with std 0.5 | trainer_tracking | 15 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"3.88\", std=\"0.25\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"0.54\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha":... |
trainer/trainer_tracking/16 | track the twist command: linear weight 1.77 (std 1.0), angular weight 2.18 (std 0.5) | trainer_tracking | 16 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"1.77\", std=\"1.0\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.18\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha": ... |
trainer/trainer_tracking/17 | velocity tracking rewards: lin 2.33/0.7071, ang 3.32/0.7071 | trainer_tracking | 17 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"2.33\", std=\"0.7071\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"3.32\", std=\"0.7071\")\ndone = b.ENO\n",
"fbd_... |
trainer/trainer_tracking/18 | reward linear velocity tracking at weight 1.13 with std 0.7071 and angular at 2.82 with std 0.5 | trainer_tracking | 18 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"1.13\", std=\"0.7071\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"2.82\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha... |
trainer/trainer_tracking/19 | track the twist command: linear weight 2.87 (std 0.25), angular weight 3.25 (std 0.5) | trainer_tracking | 19 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program tracking\nvar done : BOOL\na = CALL[Reward.track_linear_velocity](TARGET=\"reward\", weight=\"2.87\", std=\"0.25\")\nb = CALL[Reward.track_angular_velocity](EN=a.ENO, TARGET=\"reward\", weight=\"3.25\", std=\"0.5\")\ndone = b.ENO\n",
"fbd_sha":... |
trainer/trainer_gait/21 | gait shaping: air time -0.51, foot clearance 1.07 toward 0.169 m, swing height 0.9 toward 0.171 m, slip 0.44, soft landing -0.35 | trainer_gait | 21 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program gait\nvar done : BOOL\na = CALL[Reward.air_time](TARGET=\"reward\", weight=\"-0.51\")\nc = CALL[Reward.foot_clearance](EN=a.ENO, TARGET=\"reward\", weight=\"1.07\", target_height=\"0.169\")\ns = CALL[Reward.foot_swing_height](EN=c.ENO, TARGET=\"rew... |
trainer/trainer_gait/22 | set the five foot rewards: air_time 1.87; clearance -0.58 at 0.068; swing -0.93 at 0.148; slip 2.0; landing -0.45 | trainer_gait | 22 | [] | [
{
"candidate": "rank1",
"rank": 1,
"fbd_text": "program gait\nvar done : BOOL\na = CALL[Reward.air_time](TARGET=\"reward\", weight=\"1.87\")\nc = CALL[Reward.foot_clearance](EN=a.ENO, TARGET=\"reward\", weight=\"-0.58\", target_height=\"0.068\")\ns = CALL[Reward.foot_swing_height](EN=c.ENO, TARGET=\"rew... |
End of preview. Expand in Data Studio
taskweft-fbd-trainer-train
Intents paired with IEC 61131-3 Function Block Diagrams that carry them out, as an EditScore-shaped corpus.
What it is for
Each intent has a root row, three ranked candidate diagrams and a score row per candidate.
Rank1 is the reference diagram. Rank3 compiles and does the wrong thing. Rank5 is a diagram the compiler refuses.
The scores come from the trainer config: the calls were applied to the mjlab task config and the term table read back.
Every row is built from a template and a seed, so the labels are true by construction and the corpus regenerates from the seeds.
The train split trains. The test split holds out every tenth seed of the same distribution and gates the trained model. The evaluation split holds out the trainer_push family and the Reward.self_collisions block, and is never trained or tuned on.
The trainer_root config holds the intent, template, seed, frame, the block kinds and signatures the reference uses, and the input traces.
The trainer_candidates config holds each candidate's compiler text form, PLCopen XML, plan and runner result, with a sha256 of its source text.
The trainer_scores config holds parse, compile, run and effect-match flags, step and wall-time measures, and the refusal.
The default trainer config joins the three. Every artifact is in the parquet. An artifact a family does not produce is an empty string, never a null.
A manifest.json and a census.json sit at the repo root.
Provenance
- Generator:
v-sekai-fabric/taskweft-fbd-teacher. - Compiler:
taskweft-fbd-compilerat commit7d34ff4.
Use
from datasets import load_dataset
ds = load_dataset("chibifire/taskweft-fbd-trainer-train", "trainer", split="train")
Or fetch the files with hf download chibifire/taskweft-fbd-trainer-train --repo-type dataset.
Licence
MIT. See the MIT License.
- Downloads last month
- 29