# Kronos | ShiYu | Knowledge Base | Instagit

Kronos: A Foundation Model for the Language of Financial Markets

GitHub Stars: 12.4k

Repository: https://github.com/shiyu-coder/Kronos

---

## Articles

### [Computational Requirements for Kronos-base vs Kronos-mini Inference: Hardware Specifications and Performance Guide](/shiyu-coder/Kronos/computational-requirements-for-kronos-base-vs-kronos-mini-inference)

Compare computational requirements for Kronos-base vs Kronos-mini inference discover hardware specs and performance benchmarks. Kronos-mini offers faster edge deployment.

- Tags: performance
- Published: 2026-04-10

### [Key Hyperparameters for Fine-Tuning Kronos on Custom Datasets: Configuration Guide](/shiyu-coder/Kronos/key-hyperparameters-for-fine-tuning-kronos-on-custom-datasets)

Master Kronos fine-tuning with our configuration guide. Discover key hyperparameters like lookback window, predict window, epochs, and learning rates for custom datasets.

- Tags: configuration-guide
- Published: 2026-04-10

### [How to Fine-Tune the Kronos Tokenizer for New Market Data: A Complete Guide](/shiyu-coder/Kronos/how-to-fine-tune-kronos-tokenizer-for-new-market-data)

Learn to fine-tune the Kronos tokenizer for new market data. This guide details preparing CSV data, configuring YAML, and running the finetune script to adapt the Binary Spherical Quantizer.

- Tags: how-to-guide
- Published: 2026-04-10

### [Understanding the `sample_count` Parameter and Forecast Averaging in Kronos](/shiyu-coder/Kronos/purpose-of-sample-count-parameter-and-averaging-forecasts-in-kronos)

Learn how Kronos uses sample_count to average stochastic trajectories, reducing variance for stable and accurate time series forecasts.

- Tags: deep-dive
- Published: 2026-04-10

### [How Kronos Normalizes and Denormalizes Price Data During Prediction](/shiyu-coder/Kronos/how-does-kronos-normalize-and-denormalize-price-data-during-prediction)

Learn how Kronos normalizes and denormalizes price data for accurate predictions. Understand z-score transformation for time series forecasting with its per-series mean and standard deviation approach.

- Tags: deep-dive
- Published: 2026-04-10

### [BSQuantizer Entropy Penalty Terms (γ₀, γ, ζ) in Kronos Codebook Learning](/shiyu-coder/Kronos/role-of-bsquantizer-entropy-penalty-terms-in-kronos-codebook-learning)

Understand the role of BSQuantizer entropy penalty terms gamma0, gamma, and zeta in Kronos codebook learning. Control codebook diversity by regularizing entropy and overall strength.

- Tags: internals
- Published: 2026-04-10

### [How Kronos Computes Loss with DualHead for s1 and s2 Token Prediction](/shiyu-coder/Kronos/how-does-kronos-compute-loss-with-dualhead-for-s1-and-s2-token-prediction)

Learn how Kronos computes loss with DualHead for s1 and s2 token prediction by averaging cross-entropy terms and optionally masking padded positions for accurate gradient updates.

- Tags: deep-dive
- Published: 2026-04-10

### [Batch Prediction Consistency Requirements in Kronos: A Complete Guide to predict_batch()](/shiyu-coder/Kronos/requirements-for-batch-prediction-consistency-in-kronos-predict-batch)

Master Kronos batch prediction consistency with predict_batch(). Ensure identical series lengths, horizons, and OHLCV columns for accurate results. Learn the complete requirements.

- Tags: deep-dive
- Published: 2026-04-10

### [How auto_regressive_inference Manages Memory in Kronos for Long Predictions](/shiyu-coder/Kronos/how-does-auto-regressive-inference-manage-memory-in-kronos-for-long-predictions)

Master memory management in Kronos for long predictions with auto_regressive_inference. Learn how sliding windows ensure O(max_context) complexity, not prediction length. Optimize your time-series analysis.

- Tags: internals
- Published: 2026-04-10

### [What Is the clip Parameter in KronosPredictor? Default Value and Usage Explained](/shiyu-coder/Kronos/what-is-the-clip-parameter-in-kronospredictor-and-its-default-value)

Learn about the clip parameter in KronosPredictor. Understand its default value of 5 and how it prevents extreme outliers from destabilizing inference.

- Tags: api-reference
- Published: 2026-04-10

### [How Kronos TemporalEmbedding Encodes Time Information: Component-Based Temporal Encoding](/shiyu-coder/Kronos/how-does-kronos-temporalembedding-encode-time-information)

Kronos TemporalEmbedding skillfully encodes time by breaking timestamps into discrete calendar components, embedding each, and summing them for a unified temporal vector to enhance model embeddings.

- Tags: deep-dive
- Published: 2026-04-10

### [Kronos Quantization Strategy for K-line Data: A Hybrid Hierarchical-Binary Approach](/shiyu-coder/Kronos/what-quantization-strategy-does-kronos-use-for-k-line-data)

Discover Kronos' novel hybrid hierarchical-binary quantization strategy for K-line data. Learn how it transforms OHLCV time series into discrete tokens for efficient analysis.

- Tags: deep-dive
- Published: 2026-04-10

### [How Token Teacher Forcing Works in Kronos: Training and Inference Guide](/shiyu-coder/Kronos/how-does-token-teacher-forcing-work-in-kronos-training-and-inference)

Understand token teacher forcing in Kronos training and inference. Learn how the use_teacher_forcing flag controls ground-truth vs sampled embeddings for efficient model operation.

- Tags: deep-dive
- Published: 2026-04-10

### [Recommended Lookback Window Size for Kronos Predictions: Default Configuration Guide](/shiyu-coder/Kronos/recommended-lookback-window-size-for-kronos-predictions)

Discover the recommended lookback window size for Kronos predictions. Learn why the default 400 time steps is optimal for your inference needs based on the shiyu-coder/Kronos repository.

- Tags: configuration-guide
- Published: 2026-04-10

### [How Kronos Handles Missing Volume and Amount Data in Input DataFrames](/shiyu-coder/Kronos/how-does-kronos-handle-missing-volume-and-amount-data-in-input-dataframes)

Discover how Kronos addresses missing volume and amount data in input DataFrames. It automatically inserts zeros or derives data from mean price, ensuring complete OHLCVA features for model inference.

- Tags: how-to-guide
- Published: 2026-04-10

### [Kronos predict vs predict_batch: Single vs Batch Time-Series Inference Explained](/shiyu-coder/Kronos/difference-between-kronos-predict-and-predict-batch-methods)

Understand the difference between Kronos predict() and predict_batch() for time-series inference. Learn when to use single vs. batch processing for financial forecasting.

- Tags: deep-dive
- Published: 2026-04-10

### [How Temperature, Top‑k, and Top‑p Parameters Affect Sampling in Kronos Probabilistic Forecasting](/shiyu-coder/Kronos/how-do-temperature-top-k-and-top-p-parameters-affect-sampling-in-kronos-probabilistic-forecasting)

Discover how temperature, top-k, and top-p parameters influence sampling in Kronos probabilistic forecasting. Control distribution sharpness and limit vocabulary for stochastic price-volume forecasts.

- Tags: deep-dive
- Published: 2026-04-10

### [Understanding the `max_context` Parameter in KronosPredictor: Controlling Autoregressive Context Windows](/shiyu-coder/Kronos/what-is-the-max-context-parameter-in-kronospredictor-and-its-effect-on-autoregressive-generation)

Learn how the max_context parameter in KronosPredictor controls autoregressive generation by managing GPU memory and historical context for transformer decoders.

- Tags: how-to-guide
- Published: 2026-04-10

### [How Kronos Uses Rotary Positional Embedding (RoPE) for Temporal Awareness in Financial Time Series](/shiyu-coder/Kronos/how-does-kronos-use-rotary-positional-embedding-rope-for-temporal-awareness-in-financial-time-series)

Discover how Kronos leverages Rotary Positional Embedding RoPE in attention layers to achieve temporal awareness in financial time series, learning patterns without manual features.

- Tags: deep-dive
- Published: 2026-04-10

### [Kronos-Mini vs Small vs Base vs Large: Complete Model Comparison and Selection Guide](/shiyu-coder/Kronos/what-are-the-differences-between-kronos-mini-kronos-small-kronos-base-and-kronos-large-models)

Explore Kronos model differences: Mini, Small, Base, and Large. Compare parameter counts, context windows, and hardware needs to select the best fit for your project. Discover availability now.

- Tags: comparison-guide
- Published: 2026-04-10

### [How DependencyAwareLayer Enables Conditional s2 Token Generation Based on s1 Tokens in Kronos](/shiyu-coder/Kronos/how-does-dependencyawarelayer-enable-conditional-s2-token-generation-based-on-s1-tokens-in-kronos)

Learn how Kronos DependencyAwareLayer uses s1 token embeddings to conditionally generate s2 tokens via cross-attention, enhancing contextual generation.

- Tags: how-to-guide
- Published: 2026-04-10

### [What Is the Binary Spherical Quantizer (BSQuantizer) in Kronos and How Does It Compress OHLCV Data?](/shiyu-coder/Kronos/what-is-the-binary-spherical-quantizer-bsquantizer-in-kronos-and-how-does-it-compress-ohlcv-data)

Learn how the Binary Spherical Quantizer BSQuantizer in Kronos compresses OHLCV data by L2-normalizing latent vectors and quantizing them into binary codes on the unit hypersphere.

- Tags: deep-dive
- Published: 2026-04-10

### [How Kronos Tokenizes Financial Candlestick Data into s1 and s2 Tokens](/shiyu-coder/Kronos/how-does-kronos-tokenize-financial-candlestick-data-into-s1-and-s2-tokens)

Learn how Kronos tokenizes financial candlestick data into s1 and s2 tokens. Discover the Binary Spherical Quantizer and its two-level hierarchical token stream for efficient data representation.

- Tags: internals
- Published: 2026-04-10