Agent
Data Labeling The built-in assistant you build pipelines with, in a chat on the project page. You describe the task in plain language. The agent assembles the pipeline, writes the instructions and the labeling screen, and sets quality control, audience, and pricing. See How the platform works.API key
Core A credential you generate in the Business Console. It acts as your user for the HTTP API, and for the CLI when automation cannot sign in through a browser. Access follows the teams you belong to. See CLI and Programmatic access (API).Audience
Data Labeling The set of experts eligible to label on a Generation or Quorum node, filtered by language, specialization, and country. It applies to expert labeling only. Synthetic labeling has no audience. See Generation node.Audience reuse
Data Labeling Reusing the expert audience from a deployed Generation or Quorum node, typically one in another pipeline of the same project, so experts who passed that node’s onboarding start work without repeating it. It is unavailable under synthetic labeling. See Generation node.Balance
Core The prepaid funds owned by your team and shared by its projects. Pipeline runs draw down the balance as they label and pause when it no longer covers the work. A Training Run reserves its maximum charge, then settles GPU usage against that hold. The products have different settlement rules. See Billing.Base model
Train A catalog model you can train from, identified by a stable id such asqwen3-8b. A Training names either one base model or one existing Model Version as its source. See Fine-tune a model.
Batch
Data Labeling The items one run processes, up to the 10,000-item per-run limit. A large or recurring workload is a cycle of batches: deploy one, clone the pipeline, point the copy at the next, deploy again. See Label a large dataset in batches.Branch
Data Labeling One of a Match node’s outgoing paths. Each condition is a branch, labeled by the condition’s name. The default branch carries every item that matches no condition. See Match node.Business Console
Core The separate Toloka application where money enters your account: Top up and Billing history open its billing page in a new tab. The platform reads your balance from it and links to it. Adding funds, invoices, and payment details live there.Check
Data Labeling A test or lint on one node. Test checks use sample cases: Code Test, Annotation Quality Assurance, and Synthetic Generation. Built-in lints start automatically after saved edits and inspect settings or selected data. Required lints must finish before deployment. See Self-check.Checkout
Core A directory bound to one platform project by atoloka.json file. The CLI treats that file as the project root. See CLI.
Clone
Data Labeling A copy of a pipeline that carries the same configuration into a new pipeline in the same project. Because a deployed pipeline is locked, cloning is how you label another batch or start a new version. See Label a large dataset in batches.Condition
Data Labeling A named group of field tests on a Match node. Each test compares one input field’s value, and the condition passes when all of its tests pass, or any one, as configured. Each condition gets its own branch. An item that matches no condition follows the default branch. See Match node.Contract
Data Labeling The per-item prices for one run, frozen when you deploy and unchanged for the life of that run. Each run freezes its own contract, so a clone does not inherit the original’s prices. The run’s Contract button reopens it at any time. See Pricing and billing.Cost breakdown
Data Labeling The Cost breakdown (forecast) window shows the forecast before you deploy, with one row per billable line and an Expected and a Maximum total for the run. It lists the real per-task, per-call, and per-token prices and computes the totals. See Pricing and billing.Cost estimate
Train A free check of a Training’s selected rows that returns either validation problems or a maximum charge. The Training reads a frozen Dataset Slice. A ready estimate expires 24 hours after you request it. See Training.Data collection
Data Labeling A task type where experts produce new content. Uploading media is one example. Experts do not label data you already have in this task type. Its labeling interface must include a media-upload control, it is the one task type that may use the wider unverified expert pool, and synthetic labeling is not available on it. See Generation node.Dataset
Core A project-owned collection of items and fields. Data Labeling reads and writes fields through a pipeline; Train reads conversation examples from a frozen Dataset Slice and learns from selected train-role parts of its split column. See Create and manage datasets and Prepare training data from labels.Dataset Slice
Core A frozen selection of dataset rows and fields at a cutoff time. A Training uses a ready slice that includes a split column. The CLI’s--dataset option fills the dataset’s split column, creating an 80/10/10 train, validate, and test column when it has none, and freezes a slice; --slice selects a ready one. See Slices and split columns.
Deploy
Data Labeling Launching a run: you confirm the cost, the contract freezes, and the run begins labeling. Deploying locks the pipeline, so you cannot edit or rerun it and a further batch needs a clone. A deploy that failed is the exception: it frees the same pipeline to fix and deploy again. See Label a large dataset in batches.Exam
Data Labeling A short test on real tasks that experts on a Generation or Quorum node must pass before they get production tasks. Turn it on with Enable Exam on the node’s Exam card, choose a golden dataset whose rows are the exam tasks, choose how each output field is graded, and set Tasks per Expert and Pass threshold. Each expert who passes is charged once, as Exam (one-time per expert). Synthetic labeling has no exam. See Exam.Experts
Data Labeling The people who label your data. Pages also call them expert labelers. You choose the audience and set the per-task price. The agent writes the instructions and the labeling screen they see. See How the platform works.Export
Data Labeling A downloadable JSON file of a run’s results, with one entry per item carrying its inputs and the output fields the pipeline produced. Each export is a snapshot taken when you start it. Download links expire; click the download icon in the export list to generate a fresh link. See Export format for the file’s contract and Export your labeled results for the steps.Field
Core A named value on an item. Input fields are the values a node reads. Output fields are the values it writes. Each field has a name and a type. See Fields and data.Generation model
Data Labeling The model that produces the labels under synthetic labeling, chosen per node from the platform’s catalog with a recommended default. It is separate from the review model, which checks those labels. See Label with an LLM instead of experts.Gist tokens
Train The number of learned soft tokens used to represent the prompt. Set an absolute count (8, 32, or 64) or a fraction of the prompt ("1/4", "1/8", "1/16", or "1/32"). The default is "1/8". See Training settings.
Gisting
Train A Training method that learns a compact prompt representation from example conversations plus a prompt text file. It is not supervised fine-tuning. See Training.Human review
Data Labeling A review path where expert reviewers check a sample of a node’s labeled items, 10 to 100 percent of them. It is available on the Generation node only. The Quorum node and synthetic labeling do not offer it. See Quality control.Item
Core One entry in a dataset, carrying its field values. A pipeline run processes a selection of items, up to 10,000 per run. Train has separate selection limits. See Fields and data.Item filter
Data Labeling The Start node’s filter that narrows which dataset items a run launches. It matches by a field’s value, by the source that recorded a value, or by a field’s absence, and it evaluates as of the moment the run launches. See Start and End nodes.Labeling interface
Data Labeling The screen an expert fills in for one item. The agent builds the form from your task. It shows the input fields and collects a value for each output field. On the node it is the UI for Experts card, where View previews it. Use Annotate with UI in Annotation Quality Assurance to test a submission. Pages also call it the labeling screen. See Generation node.LLM review
Data Labeling A review path where an LLM checks every labeled item against your quality criteria and returns a verdict for each. It is the default review path and always covers every item, with no sampling. See Quality control.Location
Core How a part of a structured value is addressed: it starts at the value’s root, written$, and steps inward one named member or list position at a time. Match node tests, item filters, repeating-field declarations, and derived fields each use one. See Fields and data.
LoRA rank
Train A supervised fine-tuning setting: the rank of the LoRA adapter,8, 16, or 32, with 16 the default. A larger rank gives the adapter more capacity and needs more data. See Training settings.
Maximum charge
Train The most a Training Run can cost: the price of the estimate’s maximum GPU seconds, rounded up to the cent. The platform places a hold for this amount on the team balance when the run starts and charges only the GPU seconds used by a run that completes or that you cancel; a run that fails because of the platform is not charged. See Training.Model
Train A project-owned named container for immutable Model Versions produced by Training Runs. The first Training Run creates the Model from the Training name. See Fine-tune a model.Model Version
Train One immutable published result of a completed Training Run, with a unique name and a server-assigned revision. See Fine-tune a model.Node
Data Labeling One unit of a pipeline. A pipeline is a connected set of nodes, each of a small, fixed set of kinds that label, transform, or route your data. See Pipeline nodes.Offboarding rule
Data Labeling A Quorum rule on the Offboarding card, required in No QA, Offboarding Only mode. A Majority-vote agreement rule picks one single-label output field and offboards experts who persistently disagree with the majority on it. It does not grade, reject, or change answers. The mode also requires an overlap of at least 3 and the per-expert cap off. See Offboarding rules.Onboarding
Data Labeling A one-time course experts complete before they label items on a Generation or Quorum node. On the node it is the Training card, and its charge is Onboarding (one-time per expert), or Training (one-time per expert) when an exam is enabled. It is unrelated to a Train Training. It needs three to six steps and at least one quiz with correct answers. It can also include reading material, consent, practice exercises, or worked examples. Practice exercises and worked examples cannot appear in the same course. See Generation node.Overlap
Data Labeling The number of experts who independently label each item on a Quorum node — the card calls it the number of independent annotations per task. It is a whole number from 1 to 100. Raising it raises the expert cost in step. See Quorum node.Per-expert cap
Data Labeling An optional limit on how many tasks a single expert may take on a node, a whole number from 1 to 50. It is off by default. See Generation node.Per-task price
Data Labeling The price you set for one labeling task on a Generation or Quorum node. Expert Service Cost in the cost breakdown shows your charge: the per-task price plus the platform commission. On Quorum, this charge scales with the number of experts per item. See Pricing.Pipeline
Data Labeling The connected set of nodes that takes your data in and produces labeled results. The agent assembles it from your task description. You review it, launch it, and export from it. See How the platform works.Project
Core The workspace unit that holds your pipelines, datasets, secrets, Trainings, and Models. You create one by describing a task to the agent, or withtoloka project create. Every pipeline and Training belongs to one project. See Get your first labeled dataset, Fine-tune a model, and CLI.
Quality criterion
Data Labeling A named check a labeled output is judged against, with a pass condition, a fail condition, and an optional instruction for the LLM reviewer, targeting one output field. Reviewers apply the criteria and return a verdict for each. See Quality control.Reasoning effort
Data Labeling A synthetic-labeling setting for how hard the generation model reasons on each item: Low, Medium, or High on models that support it, with Medium the default. Higher effort raises the estimated tokens per item and the price. Lower effort lowers both. See Generation node.Repeating field
Data Labeling An output field that can hold several values within a single item. It returns a list rather than one value. On a Quorum node, a repeating output comes back as one list per expert. See Fields and data.Review metadata field
Data Labeling An extra output field a reviewed node adds alongside your own fields, named after the node and ending in_qa_metadata, with details about how the item’s labels were reviewed. See Quality control.
Review model
Data Labeling The model that performs LLM review on a node. It is set per node, separately from the generation model under synthetic labeling, and each model has its own per-call price. See Quality control.Review path
Data Labeling How a node’s labels are reviewed: LLM review, human review, both, or none, depending on the node. The agent turns quality control on by default. See Quality control.Run
Data Labeling One execution of a deployed pipeline over a batch of items. A run is created at deploy and does not run again. More data means more runs. See Label a large dataset in batches.Secret
Core A credential stored in your project, such as an API key, that a Code node’s environment variable can point to, so the value stays out of your code. You create one on the project’s Secrets page. The agent can reference a secret by name but cannot create one. See Code node.Self-check
Data Labeling Validation, lints, and sample tests used before deployment. Configuration errors and failed checks block deployment. Unrun test checks are warnings you can override. Required lints must finish: Quality Requirements and User Interface Types on labeling nodes, Dataset Items on Start, and Exam Golden Dataset on expert-labeling nodes with an enabled exam. See Self-check.Single-label output field
Data Labeling An output field where each expert picks exactly one label from a fixed set. On a Quorum node, agreement is measured on single-label output fields: each offboarding rule selects one. See Quorum node.Source
Core Who recorded a field’s value: a person or an automated source such as an earlier pipeline run. A field can hold several recorded values over an item’s life, each with its own source, and a node reads the latest by default. See Fields and data.Specialization
Data Labeling A skill area used to filter a node’s audience, alongside language and country. Specializations require verified experts and are not available with the unverified pool. See Generation node.Split column
Core A dataset column that assigns rows to named parts withtrain, validate, or test roles. A Training uses a split column frozen in its Dataset Slice and learns from all train-role parts by default. Use --split-column to choose the column and --parts to select particular train parts. See Slices and split columns.
Step
Data Labeling A page of an expert onboarding course. In the Items tab, Current Step identifies the node an item has reached. See Generation node.Synthetic labeling
Data Labeling Having an LLM produce a node’s labels instead of expert labelers, with each label checked against your quality criteria by a separate LLM call. Quality control is LLM review only, and the expert audience, pricing, per-expert cap, and onboarding do not apply. See Label with an LLM instead of experts.Task
Data Labeling One unit of expert work: one item shown to one expert to label or review. Pricing is per task. On a Quorum node, one item produces as many tasks as it has experts, and the per-expert cap counts tasks. See Generation node.Team
Core The account unit that owns projects and the balance. Pipeline runs and Training Runs draw on the same team’s balance. See Teams, access, and balance.toloka
Core The command-line program shipped with the platform SDK. You sign in through your browser or with an API key, select a project, reconcile pipeline source withplan and apply, and run Training commands. See CLI.
Training
Train An editable draft that names a method (supervised fine-tuning or gisting), a source model, and a frozen Dataset Slice with a split column. Training learns from the selected train-role parts;--parts narrows that selection. The CLI can prepare a slice with --dataset or use a ready slice with --slice. Updating a Training does not change an existing Training Run. See Training.