Entdecken

30 technologies shown

• RLHF technique Aligning a model with human preference using reinforcement learning on ranked outputs.
• ReAct technique Interleaving reasoning steps with tool calls in one loop.
• Residual Connection concept Adding a layer's input to its output so gradients can pass through unchanged.
• SWE-bench benchmark Real GitHub issues that a model must resolve by editing a repository.
• Sentence Transformers framework Library for embedding sentences and passages into vectors.
• Supervised Fine-Tuning technique Fine-tuning a pretrained model on demonstrations of the target behaviour.
• Temperature technique Scaling the logits to make sampling more or less random.
• The Pile dataset 800GB of text assembled from 22 diverse sources.
• AI Search application Retrieval that ranks by meaning rather than by keyword match.
• Chunking technique Splitting documents into passages small enough to retrieve and embed.
• Cross-Encoder concept Model that scores a query and a passage together in one forward pass.
• FineWeb dataset Filtered and deduplicated web corpus with a published processing pipeline.
• Function Calling technique Letting a model request a tool by name with structured arguments.
• Grounding technique Tying a model's output to supplied sources rather than its memory.
• In-Context Learning concept A model performing a task from examples in the prompt, with no weight update.
• Scaling Laws concept Empirical relationship between model size, data, compute and loss.
• Zero-Shot concept Performing a task with no examples and no training for it.
• BEIR benchmark Heterogeneous retrieval benchmark for zero-shot evaluation.
• BF16 concept Sixteen-bit float that keeps FP32's exponent range for training stability.
• BPE technique Tokenisation that builds a vocabulary by merging frequent character pairs.
• Constrained Decoding technique Restricting which tokens a model may emit at each step.
• Contrastive Learning technique Training by pulling matching pairs together and pushing others apart.
• Gemma model Google's open-weight model family, released in small sizes.
• Gradio library Builds a web interface for a model in a few lines of Python.
• Hallucination concept Fluent output that is not supported by the model's input or by fact.
• HumanEval benchmark Python programming problems scored by running the generated code.
• Image Generation application Producing images from a text description.
• LLM-as-a-Judge technique Using a capable model to score another model's output.
• LLaVA model Vision-language model built by connecting a vision encoder to a language model.
• Learning Rate concept How far each gradient step moves the weights.