Explore

30 technologies shown

• Positional Encoding concept How a transformer, which has no inherent order, learns where tokens are.
• Pretraining technique The initial large-scale training that produces a base model.
• Regularization concept Any constraint that keeps a model from fitting its training data too closely.
• Reranking technique Second-stage ordering of retrieved candidates by a more accurate scoring model.
• SQLite database Embedded relational database, used for local-first vector search.
• Self-Consistency technique Sampling several reasoning paths and taking the majority answer.
• Softmax concept Turns a vector of scores into a probability distribution.
• Speech Recognition application Transcribing spoken audio into text.
• Structured Output technique Constraining a model to emit valid JSON or a schema-conforming object.
• Synthetic Data technique Training examples generated rather than collected.
• ViT architecture Transformer that treats an image as a sequence of patches.
• Chain of Thought technique Prompting a model to reason step by step before answering.
• DPO technique Aligning a model directly on preference pairs, without training a reward model.
• Query Rewriting technique Reformulating a question before retrieval to improve what comes back.
• Agent Memory technique State an agent keeps beyond the current context window.
• CNN architecture Convolutional neural network, sharing weights across positions to exploit spatial structure.
• FP8 concept Eight-bit floating point, used to train and serve models faster.
• GPQA benchmark Graduate-level science questions written to resist search and recall.
• HyDE technique Embedding a generated answer instead of the question.
• Keras framework High-level modelling API that runs on several backends.
• Layer Normalization concept Normalising activations across features within each sample.
• Logits concept The raw, unnormalised scores a model produces before softmax.
• MT-Bench benchmark Multi-turn conversation quality judged by a model acting as judge.
• Ray framework Distributed execution framework for scaling Python workloads.
• Recommendation application Suggesting items a user is likely to want, from behaviour and content.
• Redis database In-memory data store whose vector search runs in the same place as the cache.
• Safetensors infrastructure File format for model weights designed to avoid arbitrary code execution.
• TRL library Training library for fine-tuning and aligning models on preference data.
• Token concept The unit of text a language model reads and writes.
• Tree of Thoughts technique Searching over intermediate reasoning states instead of one linear chain.