Nouvelles pages
Aller à la navigation
Aller à la recherche
- 19 septembre 2023 à 09:33 AI Agent (hist | modifier) [208 octets] Jboscher (discussion | contributions) (Page créée avec « AI agents are artificial entities that sense their environment, make decisions, and take actions. The Rise and Potential of Large Language Model Based Agents: A Survey : https://arxiv.org/pdf/2309.07864.pdf »)
- 1 août 2023 à 12:00 SFT (hist | modifier) [1 113 octets] Jboscher (discussion | contributions) (Page créée avec « Supervised Fine-Tuning (SFT): Models are trained on a dataset of instructions and responses. It adjusts the weights in the LLM to minimize the difference between the generated answers and ground-truth responses, acting as labels. == Références == * [https://towardsdatascience.com/fine-tune-your-own-llama-2-model-in-a-colab-notebook-df9823a04a32] Fine-Tune Your Own Llama 2 Model in a Colab Notebook »)
- 27 juillet 2023 à 09:34 PEFT (hist | modifier) [373 octets] Jboscher (discussion | contributions) (Page créée avec « Thus, we, as a com- munity of researchers and engineers, need efficient ways to train on downstream task data. Parameter-efficient fine-tuning, which we denote as PEFT, aims to resolve this problem by only training a small set of parameters which might be a subset of the existing model parameters or a set of newly added parameters. https://arxiv.org/pdf/2303.15647.pdf »)
- 5 juillet 2023 à 21:18 Agent (hist | modifier) [1 406 octets] Jboscher (discussion | contributions) (Page créée avec « In LangChain, agents are high-level components that use language models (LLMs) to determine which actions to take and in what order. An action can either be using a tool and observing its output or returning it to the user. Tools are functions that perform specific duties, such as Google Search, database lookups, or Python REPL. Agents involve an LLM making decisions about which Actions to take, taking that Action, seeing an Observation, and repeating that until... »)
- 4 juillet 2023 à 09:20 Prompt design (hist | modifier) [1 029 octets] Jboscher (discussion | contributions) (Page créée avec « Please note a few best practices around prompt design. Be concise Be specific and well-defined Ask one task at a time Turn generative tasks into classification tasks. For example, instead of asking what programming language to learn, ask if Python, Java, or C is a better fit for a beginner in programming. and Improve response quality by including examples. Adding instructions and a few examples tends to yield good results however there’s currently no one best w... »)
- 4 juillet 2023 à 08:58 Zero-shot (hist | modifier) [2 799 octets] Jboscher (discussion | contributions) (Page créée avec « Zero-shot prompting - is a method where the LLM is given no additional data on the specifictask that it is being asked to perform. Instead, it is only given a prompt that describes the task. For example, if you want the LLM to answer a question, you just prompt "what is prompt design?". One-shot prompting - is a method where the LLM is given a single example of the task that it is being asked to perform. For example, if you want the LLM to write a poem, you might... »)
- 22 juin 2023 à 08:30 Top P (hist | modifier) [3 088 octets] Jboscher (discussion | contributions) (Page créée avec « First, there are different models you can choose from. Each model is tuned to perform well on specific tasks. You can also specify the temperature, top P, and top K. These parameters all adjust the randomness of responses by controlling how the output tokens are selected. When you send a prompt to the model, it produces an array of probabilities over the words that could come next. And from this array, we need some strategy to decide what to return. A simple stra... »)
- 22 juin 2023 à 08:28 Top K (hist | modifier) [2 291 octets] Jboscher (discussion | contributions) (Page créée avec « First, there are different models you can choose from. Each model is tuned to perform well on specific tasks. You can also specify the temperature, top P, and top K. These parameters all adjust the randomness of responses by controlling how the output tokens are selected. When you send a prompt to the model, it produces an array of probabilities over the words that could come next. And from this array, we need some strategy to decide what to return. A simple stra... »)
- 22 juin 2023 à 08:24 Temperature (hist | modifier) [2 013 octets] Jboscher (discussion | contributions) (Page créée avec « First, there are different models you can choose from. Each model is tuned to perform well on specific tasks. You can also specify the temperature, top P, and top K. These parameters all adjust the randomness of responses by controlling how the output tokens are selected. When you send a prompt to the model, it produces an array of probabilities over the words that could come next. And from this array, we need some strategy to decide what to return. A simple stra... »)
- 11 mai 2023 à 03:30 LangChain (hist | modifier) [935 octets] Jboscher (discussion | contributions) (Page créée avec « == Références == * [https://python.langchain.com/en/latest/index.html] Documentation LangChain * [https://www.youtube.com/watch?v=2xxziIWmaSA&list=PLqZXAkvF1bPNQER9mLmDbntNfSpzdDIU5&index=4] Cours LangChain »)
- 2 mai 2023 à 09:25 LLaMA (hist | modifier) [337 octets] Jboscher (discussion | contributions) (Page créée avec « LLaMA (Large Language Model Meta AI) is a language model released by Meta (Facebook). It is Meta’s answer to OpenAI’s GPT models. Like GPT, LLaMA is intended to be a general-purpose foundational model suitable for further fine-tuning. [1] == Références == * [https://agi-sphere.com/llama-models/] A brief history of LLaMA models »)
- 1 mai 2023 à 07:29 Prompt (hist | modifier) [3 863 octets] Jboscher (discussion | contributions) (Page créée avec « == Références == * [https://learn.microsoft.com/en-us/azure/cognitive-services/openai/concepts/prompt-engineering] Introduction to prompt engineering * [https://www.deeplearning.ai/short-courses/chatgpt-prompt-engineering-for-developers/] ChatGPT Prompt Engineering for Developers * [https://learn.microsoft.com/en-us/azure/cognitive-services/openai/concepts/advanced-prompt-engineering?pivots=programming-language-chat-completions] Prompt engineering techniques... »)
- 27 avril 2023 à 16:35 In-context learning (hist | modifier) [571 octets] Jboscher (discussion | contributions) (Page créée avec « [1] The in-context learning (ICL) ability is formally introduced by GPT-3 : assuming that the language model has been provided with a natural language instruction and/or several task demonstrations, it can generate the expected output for the test instances by completing the word sequence of input text, without requiring additional training or gradient update == Références == * [https://arxiv.org/pdf/2303.18223.pdf] A Survey of Large Language Models »)
- 27 avril 2023 à 15:39 Emergent Abilities of Large Language Models (hist | modifier) [1 583 octets] Jboscher (discussion | contributions) (Page créée avec « On entend par "Emergent Abilities of Large Language Models" une capacité présente dans un LLM qui ne se retrouve pas dans un modèle similaire mais plus petit. Ce qui veut dire aussi qu'on ne peut pas prévoir (extrapoler) cette nouvelle capacité uniquement à partir de celles d'un modèle plus petit. [1] We consider an ability to be emergent if it is not present in smaller models but is present in larger models. Although scaling is mainly conducted in mode... »)
- 27 avril 2023 à 14:44 Pre-trained language models (hist | modifier) [1 306 octets] Jboscher (discussion | contributions) (Page créée avec « Les auteurs de "A Survey of Large Language Models" distinguent les "Pre-trained language models (PLM)" des Large language models (LLM). ELMo et BERT appartiendraient à la 1ère catégorie, un peu comme les pionniers des LLM. As an early attempt, ELMo was proposed to capture context-aware word representations by first pre-training a bidirectional LSTM (biLSTM) network (instead of learning fixed word representations) and then fine-tuning the biLSTM network acco... »)