Datasets:

mkurman
/

agentic-synth-reasoning

messages listlengths 4 24	tools stringlengths 3.92k 15.9k	domain stringclasses 457 values	source_dataset stringclasses 1 value	original_model stringclasses 1 value	num_original_messages int64 5 25	num_reasoning_turns int64 2 8	generated_at stringlengths 32 32
[ { "role": "user", "content": "Hi, I'm reaching out to learn more about the regenerative farming practices you use. I'm particularly interested in how these methods impact soil health, especially regarding the recent delivery I received. Could you provide more details on your sustainable practices?", "to...	[{"type": "function", "function": {"name": "verify_user_identity", "description": "Authenticate registered users via email/phone/account number with verification code.", "parameters": {"properties": {"user_id": {"type": "string"}, "auth_method": {"enum": ["email", "phone", "account_number"], "type": "string"}, "verific...	agriculture consumer-facing	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	5	2	2026-07-04T16:55:49.744059+00:00
[ { "role": "user", "content": "Hello, this is Officer Mark Thompson from the Department of Food Safety Inspection. I'm calling from the government number 202-555-0198, and my office is based in ZIP code 20005. I need to request comprehensive audit documentation for farms under your organization's purview. Sp...	[{"type": "function", "function": {"name": "authenticate_user", "description": "Authenticate a subscriber using registered email or phone and ZIP code", "parameters": {"type": "object", "properties": {"identifier": {"type": "string"}, "zip_code": {"type": "string"}}, "required": ["identifier", "zip_code"]}, "strict": t...	agriculture consumer-facing	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	9	4	2026-07-04T16:55:53.497441+00:00
[ { "role": "user", "content": "Hello, I'm reaching out from a 501(c)(3) nonprofit organization, and we have a collection of 15 antiques valued at $22,000 that we'd like to consign with your auction house. Given our nonprofit status, I was wondering if you might consider offering a discounted commission rate ...	[{"type": "function", "function": {"name": "get_auction_details", "description": "Retrieve auction event details including closing times and items", "parameters": {"properties": {"auction_event_id": {"description": "Unique identifier for the auction event", "title": "Auction Event Id", "type": "string"}}, "required": [...	antique auction houses	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	9	4	2026-07-04T16:55:58.927084+00:00
[ { "role": "user", "content": "Hi, I'm looking for a dispensary hat as a birthday gift. Do you have any branded merchandise available?", "tool_call_id": "", "reasoning_content": null }, { "role": "assistant", "content": "I'm here to help with our cannabis products, but I need to let you k...	[{"type": "function", "function": {"name": "verify_age", "description": "Verify customer age and medical status using government ID scan and optional medical authorization document. Required before discussing products or accessing customer data.", "parameters": {"properties": {"customer_id": {"description": "Unique cus...	cannabis retail dispensaries	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	7	3	2026-07-04T16:56:08.313739+00:00
[ { "role": "user", "content": "Hello, I'm the logistics director for a premium-tier aerospace manufacturer (Client ID: AM-4488). I need immediate assistance regarding Booking ID BK776655 for aircraft components en route from Hamburg to Seattle. The Port of Antwerp has been closed due to a strike, triggering ...	[{"type": "function", "function": {"name": "authenticate_client", "description": "Verify client identity using either: 1) client_id and booking_id OR 2) company_affiliation and auth_token as per authentication protocol.", "parameters": {"properties": {"client_id": {"description": "Unique client identifier", "type": "st...	maritime shipping brokers	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	5	2	2026-07-04T16:56:10.070025+00:00
[ { "role": "user", "content": "Hello, I need assistance with an issue regarding my recent order. I received a crushed cereal box (Order #FM2025-5550) in unopened condition. It was delivered less than an hour ago. Can you process a refund per your non-perishable damage policy?", "tool_call_id": "", "r...	[{"type": "function", "function": {"name": "verify_user_identity", "description": "Verify user identity using valid credentials", "parameters": {"properties": {"method": {"enum": ["customer_id_pin", "phone_code"], "description": "Authentication method to use", "type": "string"}, "customer_id": {"description": "Unique c...	grocery stores	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	19	8	2026-07-04T16:56:14.326924+00:00
[ { "role": "user", "content": "Hello, I'm calling regarding my order SRT-2024-333 for trousers. I need to discuss a waist reduction from 40\" to 35\"—would this 5\" alteration be possible? I understand there might be a ±3cm boundary, but I’d appreciate your guidance on next steps.", "tool_call_id": "", ...	[{"type": "function", "function": {"name": "get_customer_profile", "description": "Retrieve customer details including measurements, contact preferences, payment methods, and order history.", "parameters": {"type": "object", "properties": {"customer_id": {"type": "string", "description": "Unique customer identifier"}},...	bespoke tailoring houses	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	5	2	2026-07-04T16:56:25.337760+00:00
[ { "role": "user", "content": "Hello, my Customer ID is SG990044 and my PIN is 5872. I’m reaching out to inquire about expanding my security monitoring to the detached pool house at my property located at 202 Pool Property. The pool house is 400 sq ft. Could you let me know what options are available for cov...	[{"type": "function", "function": {"name": "authenticate_customer", "description": "Authenticate customer using primary or secondary methods.", "parameters": {"type": "object", "properties": {"customer_id": {"description": "Unique customer identifier", "type": "string"}, "authentication_method": {"enum": ["pin", "phone...	security system companies	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	13	6	2026-07-04T16:56:32.267152+00:00
[{"role":"user","content":"Hello, I'm a commercial ferry maintenance supervisor contacting regarding(...TRUNCATED)	"[{\"type\": \"function\", \"function\": {\"name\": \"get_customer_profile\", \"description\": \"Ret(...TRUNCATED)	marine equipment suppliers	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	13	6	2026-07-04T16:56:33.343388+00:00
[{"role":"user","content":"Hello, I'm a military family member with a dependent card from the Army. (...TRUNCATED)	"[{\"type\": \"function\", \"function\": {\"name\": \"send_verification_code\", \"description\": \"S(...TRUNCATED)	optical retail chains	nvidia/Nemotron-SFT-Agentic-v2	deepseek/DeepSeek-V3.2	5	2	2026-07-04T16:56:33.945200+00:00

End of preview. Expand in Data Studio

Agentic Synth Reasoning

Synthetic reasoning traces for multi-step agentic behavior, generated incrementally from nvidia/Nemotron-SFT-Agentic-v2.

Dataset Summary

Each record contains a multi-turn agentic conversation with synthetically generated reasoning traces (reasoning_content) attached to each assistant turn. The reasoning was produced incrementally — for each assistant message, only the conversation prefix up to that point was fed to the reasoning model, ensuring zero future leakage.

3,615 records across 457 unique domains
17,882 reasoning turns (99.9% structured format)
Average 3,002 chars per reasoning trace
Tool definitions from the source dataset included in reasoning context

Generation Method

Source: Streamed conversations from interactive_agent, tool_calling, and search splits of Nemotron-SFT-Agentic-v2
Incremental reasoning: For each assistant message at position i, only messages[0..i-1] + tool definitions were fed to the model — making future leakage physically impossible
Structured format: Reasoning follows a ### Situation Assessment → ### Decision Rationale structure with ● (confirmed), ◐ (inferred), ○ (uncertain) confidence markers
Temperature escalation retry: Empty/None responses retried at increasing temperatures (0.7 → 0.85 → 1.0)
Models used: GLM-5.2 (via Ollama), DeepSeek V4 Flash (via OpenRouter)

Data Format

{
    "messages": [
        {"role": "system", "content": "..."},
        {"role": "user", "content": "..."},
        {"role": "assistant", "content": "...", "reasoning_content": "### Situation Assessment\n● ..."},
        {"role": "tool", "content": "...", "tool_call_id": "..."},
        ...
    ],
    "domain": "theme park attraction management",
    "source_dataset": "nvidia/Nemotron-SFT-Agentic-v2",
    "original_model": "deepseek/DeepSeek-V3.2",
    "num_original_messages": 12,
    "num_reasoning_turns": 5,
    "generated_at": "2026-07-04T16:55:49.744059+00:00"
}

Validation

Check	Result
Format compliance	99.9% (17,864/17,882 turns)
Future leakage	0 confirmed (incremental generation)
Corrupted fields	0
Duplicates	0

Intended Use

Designed as an SFT warmup dataset for GRPO-based LLM evaluator training. The reasoning traces teach models to:

Decompose user requests in agentic contexts
Select appropriate tools from available options
Justify decisions with policy/constraint awareness
Assess risks and edge cases before acting

Source Data

Derived from nvidia/Nemotron-SFT-Agentic-v2. Original conversations preserved; only reasoning_content fields are synthetically generated.

Downloads last month: 31