messages listlengths 17 236 | tools stringclasses 1
value | source stringclasses 115
values | metadata dict |
|---|---|---|---|
[
{
"role": "system",
"content": "You are an analytics engineer working inside a DuckDB/dbt project.\nComplete the requested analytics-engineering task in the existing project. Inspect the relevant project files and data, implement the requested analytical result, run the necessary dbt commands, inspect execu... | [{"type": "function", "function": {"name": "bash", "description": "Run a bash command. Each call starts a fresh shell in the project directory. Returns exit code and output.", "parameters": {"type": "object", "properties": {"command": {"type": "string"}}, "required": ["command"]}}}, {"type": "function", "function": {"n... | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {
"benchmark": "ade",
"task": "ade-bench-airbnb005",
"attempt": 0,
"reward": 0,
"success": false,
"stop": "finished",
"num_turns": 26,
"seconds": 177,
"model": "Qwen/Qwen3.5-35B-A3B",
"enable_thinking": false,
"max_turns": 100,
"error": null
} |
[
{
"role": "system",
"content": "You are an analytics engineer working inside a DuckDB/dbt project.\nComplete the requested analytics-engineering task in the existing project. Inspect the relevant project files and data, implement the requested analytical result, run the necessary dbt commands, inspect execu... | [{"type": "function", "function": {"name": "bash", "description": "Run a bash command. Each call starts a fresh shell in the project directory. Returns exit code and output.", "parameters": {"type": "object", "properties": {"command": {"type": "string"}}, "required": ["command"]}}}, {"type": "function", "function": {"n... | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {
"benchmark": "ade",
"task": "ade-bench-airbnb005",
"attempt": 1,
"reward": 0,
"success": false,
"stop": "finished",
"num_turns": 22,
"seconds": 134,
"model": "Qwen/Qwen3.5-35B-A3B",
"enable_thinking": false,
"max_turns": 100,
"error": null
} |
[
{
"role": "system",
"content": "You are an analytics engineer working inside a DuckDB/dbt project.\nComplete the requested analytics-engineering task in the existing project. Inspect the relevant project files and data, implement the requested analytical result, run the necessary dbt commands, inspect execu... | [{"type": "function", "function": {"name": "bash", "description": "Run a bash command. Each call starts a fresh shell in the project directory. Returns exit code and output.", "parameters": {"type": "object", "properties": {"command": {"type": "string"}}, "required": ["command"]}}}, {"type": "function", "function": {"n... | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {
"benchmark": "ade",
"task": "ade-bench-airbnb005",
"attempt": 10,
"reward": 0,
"success": false,
"stop": "finished",
"num_turns": 24,
"seconds": 115,
"model": "Qwen/Qwen3.5-35B-A3B",
"enable_thinking": false,
"max_turns": 100,
"error": null
} |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":11,"reward":0.0,"success":false,"stop":"fi(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":12,"reward":1.0,"success":true,"stop":"fin(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":13,"reward":0.0,"success":false,"stop":"fi(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":14,"reward":0.0,"success":false,"stop":"fi(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":15,"reward":1.0,"success":true,"stop":"fin(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":16,"reward":1.0,"success":true,"stop":"fin(...TRUNCATED) |
[{"role":"system","content":"You are an analytics engineer working inside a DuckDB/dbt project.\nCom(...TRUNCATED) | "[{\"type\": \"function\", \"function\": {\"name\": \"bash\", \"description\": \"Run a bash command.(...TRUNCATED) | qwen3.5-35b-a3b_self/ade/ade-bench-airbnb005 | {"benchmark":"ade","task":"ade-bench-airbnb005","attempt":17,"reward":1.0,"success":true,"stop":"fin(...TRUNCATED) |
End of preview. Expand in Data Studio
dbt agent SFT trajectories
Agent trajectories of Qwen3.5-35B-A3B (no thinking) solving dbt / SQL data-engineering tasks, collected for
self-distillation in the INF 385T LLM post-training course project. Layout follows
allenai/tmax-sft: messages, tools, source, metadata.
| Subset | Rows | Tasks |
|---|---|---|
qwen3.5-35b-a3b_self_all |
7,475 | 115 |
qwen3.5-35b-a3b_self_only_success |
1,383 | 72 |
- Tasks: 115 train tasks (55 Spider2-DBT, 16 ADE-bench, 44 data-eng-bench analytics); the 20 held-out data-eng-bench test tasks of the v2 split are excluded (10 of the 44 data-eng-bench tasks here are test tasks in the later v3 split).
- Rollouts: 65 attempts per task, at most 100 turns, served with vLLM (FP8). Reward 1 = the task's tests pass.
- Tools:
bash,read_file,write_file,edit_file(OpenAI function-calling format, JSON intools). _only_success: rollouts with reward 1, the self-distillation SFT data. Training used only examples of at most 49,152 tokens under the Qwen3.5 chat template.metadata: benchmark, task, attempt, reward, success, stop reason, number of turns, wall-clock seconds, model.- Twilio account SIDs appearing in tool output are replaced by
AC_REDACTED_SID.
from datasets import load_dataset
ds = load_dataset("xliaoyi/dbt-agent-sft", "qwen3.5-35b-a3b_self_only_success", split="train")
Code and evaluation results: https://github.com/xliaoyi/llms-post-train
- Downloads last month
- 147