Ecom-RLVE：Eコマース対話エージェントのための適応可能な検証環境

Back to Articles Ecom-RLVE: Adaptive Verifiable Environments for E-Commerce Conversational Agents

Upvote - Rahul Bajaj thebajajra Follow owlgebra-ai Jaya Nupur ai-queen Follow owlgebra-ai Anuj Garg pmonad Follow owlgebra-ai ben burtenshaw burtenshaw Follow TL;DR — We extend the RLVE framework from single-turn reasoning puzzles to multi-turn, tool-augmented e-commerce conversations. EcomRLVE-GYM provides 8 verifiable environments — product discovery, substitution, cart building, returns, order tracking, policy QA, bundle planning, and multi-intent journeys — each with procedural problem generation, a 12-axis difficulty curriculum, and algorithmically verifiable rewards. We train a Qwen 3 8B model with DAPO over 300 steps and present early results demonstrating that environment scaling and adaptive difficulty transfer to agentic, real-world task completion.

This project originated in the Pytorch OpenEnv Hackathon and is still evolving, follow us for updates 🔥

Why RL for shopping agents?

Large language models can hold fluent conversations, yet deploying them as shopping assistants reveals a persistent gap: fluency ≠ task completion. A customer who asks "find me a USB-C charger under $25 that ships in two days" needs an agent that invokes the right catalog search, filters on three hard constraints, avoids hallucinating product IDs it never retrieved, and handles follow-ups when the top result goes out of stock.

Supervised fine-tuning can teach surface-level tool use from demonstrations, but it cannot scale to the combinatorial space of constraint configurations, partial-information dialogues, and multi-step transactional workflows that real e-commerce demands.

Reinforcement learning with verifiable rewards (RLVR) offers an alternative: the agent optimises for outcomes — did the products satisfy the constraints? Was the cart correct? Was the return initiated for the right order line? The challenge is constructing reward functions that are both verifiable (no LLM-as-a-judge subjectivity) and adaptive (difficulty that grows with the policy's capability).

From RLVE-Gym to EcomRLVE-GYM

RLVE-Gym provides 400 environments for sorting, multiplication, Sudoku, and other algorithmic-reasoning tasks; however, those are all single-turn, text-in / text-out puzzles — extending to agentic domains was left as future work.

EcomRLVE-GYM fills that gap: we stay in the verifiable regime (e-commerce outcomes can be checked algorithmically) while extending to multi-turn, tool-augmented, agentic conversations — environments where the agent must act (call tools, modify world state) rather than merely reason (produce a text answer) and compensates for the deficiency of the search system.

EcomRLVE-GYM transforms customer-service outcomes structurally verifiable:

Every signal above can be evaluated by a program with access to the hidden ground-truth goal. No human annotation or LLM-as-a-judge is needed.

What a training episode looks like

Before we explain the framework, here is what a single EcomRLVE episode looks like at difficulty d = 4

The reward is fully computed by code: F1 over (product, variant, qty)

The eight environments

Each environment covers a distinct real-world shopping scenario. The agent must complete the task using tools (catalog search, cart operations, order lookups, policy queries) and is scored by a program — not a human or another LLM.

What the agent must do

Product Discovery

Find products that satisfy all the user's constraints

An item is out of stock — find a similar, compatible alternative

Add the exact products, variants, and quantities the user asked for

Return + Replacement

Identify the right order line, initiate a return, suggest a replacement

Resolve which order the user means and report its current status

Answer a deterministic question about store policy (return window, shipping rules, etc.)

Bundle Planning

Recommend a complete shopping list for a project within a budget

Multi-Intent Journey

Handle a conversation that chains 2–5 of the above tasks in sequence

Every environment uses the same three-part reward signal:

Task reward — did the agent actually complete the goal? (e.g., were the right products recommended, was the cart correct, was the right order tracked?)

Efficiency reward — did the agent complete it without wasting turns? Turns the user caused (asking a follow-up, confirming an action) don't count against the agent — only turns caused by agent mistakes do.

Hallucination penalty — did the agent only recommend products it actually retrieved during the session? Recommending product IDs that were never looked up is penalised, so the agent cannot invent results from memory.

Invalid outputs (malformed JSON, illegal tool calls) trigger an immediate failure score, creating a strong incentive for well-formed responses from step one.

Adaptive difficulty curriculum

A single difficulty number d

Here are four representative difficulty axes:

How many constraints the user has

How often the user omits a constraint

Fraction of search results that are distractors

Items that go out of stock mid-conversation

The other eight axes cover turn budget, input noise (typos, slang), context switches, retrieval depth, order-history size, policy complexity, and tool budget. The full breakdown is in the technical report.

Adaptive scheduling. Each environment tracks the agent's success rate independently and only advances to harder problems once the agent is passing the current level reliably. This keeps every environment training at the agent's capability frontier — avoiding both "too easy to learn from" and "too hard to make progress on".

Deep dive: Cart Building (E_CART)

Cart building is a good showcase because it requires the full search → inspect → clarify → act loop, has a binary ground truth, and introduces a challenge absent from most recommendation benchmarks: variant selection.

To succeed, the agent must develop five distinct skills:

What it means in practice

Product Discovery

Search the catalog with well-formed queries to find the right items

Variant Selection

Identify the correct color, size, or connector type — not just the right product

Cart Management

Add items with the exact variant and quantity the user asked for

Clarification Dialogue

Ask the user a focused follow-up when a request is ambiguous (e.g., missing size)

Multi-Item Orders

Handle shopping lists with several different products in a single conversation

The agent uses six tools to accomplish this:

Searches the product catalog with a natural-language query

catalog_get_variants

Returns available variants (color, size, connector, etc.) for a product

Adds a product to the cart with a specific variant and quantity

Reads the current cart so the agent can verify it matches the request

user_get_visit_history

Fetches recently viewed products by user

Sends a clarification question to the customer when a detail is missing

The generator samples 1–5 target products (scaling in difficulty with d

Search the catalog to find each product

Call catalog.get_variants

Add the correct (product_id, variant_id, qty)

Why variants matter

Real product catalogs have sparse variant data — many products have none, and those that do typically vary only by colour or size. To create a richer discrimination task, we synthesize variants at episode initialization:

A per-category priority list picks the most natural attribute to vary (electronics → connector_type

For each target product, we generate 3 variants: 1 target + 2 plausible distractors. An "Anker 65W USB-C Charger" produces {USB-C, Lightning, HDMI}

The verifier checks composite keys (product_id, variant_id)

Difficulty scaling

Variant required

catalog.search → cart.add

The cart must be exactly right — correct product, correct variant, correct quantity. Partial credit is given for partially correct carts, but a perfect score requires every item to match. If the agent adds the wrong variant, the simulated user corrects it mid-dialogue ("that's the Lightning version, but I need USB-C"), giving the agent a chance to self-correct before the episode ends.

Trajectories: easy vs. hard

Two real E_CART episodes from a Qwen 3 8B agent. Same environment, same agent — difficulty alone changes the game.

d = 1 — 1 item, no variants

d = 8 — 3 items, variants + typos

"Shopping list: SonicWave Keypad. Please add everything."

"pls add the following to my aart: Eco-Frienly HrenLefa Charger by GreenLeaf, High-Performance AqaPure Air Fryer , Lighrweiht PixelVault Tank Top (3)." — typos from noise axis

user.get_visit_history()

user.get_visit_history()

cart.add("syn_000162", qty=1)

No tool calls — presents 3 product cards. User: "Add the Charger with 100W output, Air Fryer with Charcoal filter, 3 Tank Tops in size XS."

answer: { "done": true }

catalog.get_variants

cart.add(charger, "100W", qty=1)

cart.add(fryer, "Bamboo", qty=1)

cart.add(tank_top, "size_XL", qty=3)

Re-fetches tank top variants, adds 2× XS — doesn't fix the air fryer. User: "Air Fryer still shows Bamboo. I need three XS, not two."

Re-fetches tank top variants again — still ignores the air fryer. User: "Please change the Air Fryer variant."

"I cannot find a Charcoal filter variant." — It exists; the agent saw it in turn 3 but never selected it. User: "Skip the Air Fryer then."

answer: { "done": true }

Cart matches goal. 3 turns, 2 effective.

Wrong variants, wrong quantities, user gave up. 8 turns, 6 effective.

At d=1 the agent solves the task in 3 clean turns. At d=8 it spirals — picking Bamboo instead of Charcoal, XL instead of XS, never fixing the air fryer despite two user corrections, then hallucinating that the variant doesn't exist. This is exactly the kind of multi-step error cascade that the difficulty curriculum surfaces, and that adaptive training should teach the agent to recover from.

User simulation

A verifiable environment needs a user simulator that behaves realistically. We use Qwen3.5 (9.7B) to generate natural, varied user messages rather than canned templates — covering everything from typo-filled requests to mid-conversation topic switches.

Two design choices matter for training quality:

Preferences match stated constraints. Each simulated user has a hidden set of preferences (price sensitivity, brand loyalty, shipping speed, etc.). These are deliberately biased toward whatever constraints the user communicated — so if the user said "under $25", the reward function actually cares about price. Without this, an agent could be penalised for correctly following the user's instructions.

Strategic omission. The LLM deliberately withholds some constraints from the opening message to force the agent to ask clarifying questions. The system tracks exactly what was and wasn't mentioned, so the agent is never penalised for information it was never given.

Environment scaling

Following RLVE's methodology, we define nested environment collections:

C1 ⊂ C2 ⊂ C4 ⊂ C8

Serarch Query Formulation, Cart Manipulation

Similarity reasoning under constraints

+ Product Discovery, Returns

Transactional workflows (Retrieval + recommendation, return initiation)

+ Status, Policy, Bundle, Journey

Knowledge retrieval, planning, compositionality

We hypothesise — consistent with RLVE's findings — that C8 agents outperform single-environment specialists, even on the specialist's own task.

We trained Qwen 3 8B with DAPO on C1 (Cart Building) for 300 steps as an initial viability study.

DAPO (G = 8 rollouts/prompt)

2M products, FAISS index with Alibaba-NLP/gte-modernbert-base

We saw progressive growth in difficulty reached, confirming that adaptive scheduling produces a steady learning signal rather than the saturation (static-low) or starvation (static-high) patterns predicted by the RLVE paper.

Try it yourself

Run a live episode directly in your browser using the embedded demo below. Here is how to get started:

Pick an environment from the dropdown (e.g., E_CART

Set a difficulty — 0

Click "Reset Episode" — the simulated user will open with a shopping request.

You are the agentnow: Make tool calls, analyse outputs and submit the final list of product ids.

Click "Reset Episode" between runs to start a fresh scenario.

The environments, verifiers, and training configs are all open-source:

git clone https://github.com/owlgebra-ai/EcomRLVE-Gym cd EcomRLVE-Gym pip install -e .

The 2M-product catalog is on the Hub:

from datasets import load_dataset catalog = load_dataset("owlgebra-ai/Amazebay-catalog-2M", split="train") print(f"{len(catalog)} products loaded")

Zeng, Z., Ivison, H., Wang, Y., et al. (2025). RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments. ICML 2025. arXiv:2511.07317

Yu, Q., Zhang, Z., Zhu, R., et al. (2025). DAPO: An Open-Source LLM Reinforcement Learning System at Scale. arXiv:2503.14476

Shao, Z., Wang, P., Zhu, Q., et al. (2024). DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models. arXiv:2402.03300

DeepSeek-AI. (2025). DeepSeek-R1: Incentivizing Reasoning in LLMs through Reinforcement Learning. Nature.

Meta AI. (2024). Llama 3.1: A Foundation Model for General Intelligence. llama.meta.com

Qwen Team. (2025). Qwen3 Technical Report. arXiv:2505.09388