公开(公告)号：US11763083B2

公开(公告)日：2023-09-19

申请号：US17798638

申请日：2020-05-18

Applicant: Google LLC

Inventor： Xinying Song ,  Yang Song

IPC: G06F40/30 ,  G06F40/284 ,  G06F16/31 ,  G06F40/40

CPC classification number: G06F40/284 ,  G06F16/322 ,  G06F40/40

Abstract: Systems and methods for performing inference for word or wordpiece tokenization are disclosed using a left-to-right longest-match-first greedy process. In some examples, the vocabulary may be organized into a trie structure in which each node includes a precomputed token or token ID and a fail link, so that the tokenizer can parse the trie in a single pass to generate a list of only those tokens or token IDs that correspond to the longest matching vocabulary entries in the sample string, without the need for backtracking. In some examples, the vocabulary may be organized into a trie in which each node has a fail link, and any node that would share token(s) or token_ID(s) of a preceding node is instead given a prev_match link that points back to a chain of nodes with those token(s) or token_ID(s).

公开(公告)号：US20240289395A1

公开(公告)日：2024-08-29

申请号：US18528142

申请日：2023-12-04

Applicant: Google LLC

Inventor： Hao Zhou ,  Shrestha Basu Mallick ,  Trevor Strohman ,  Patricia Luisa Romero Domingo ,  Amirhossein Kiani ,  Yu Du ,  Xinying Song ,  Heng-Tze Cheng ,  Quoc V. Le ,  Ed Huai-Hsin Chi ,  Christopher Jamie Maclean Hall

IPC: G06F16/9532 ,  G06F16/955 ,  G06F40/40

CPC classification number: G06F16/9532 ,  G06F16/955 ,  G06F40/40

Abstract: Implementations relate to helping a large language model generate factual responses to prompts that request factual content is disclosed. The large language model may receive a prompt context, a plurality of encoded context passages as input. The large language model is trained to determine whether or not to utilize the encoded context passages in generating the response. Implementations also relate to different methods of fine-tuning the responses generated by the large language model through query refinements, response re-writes, and evaluation of factual accuracy.

公开(公告)号：US20240054288A1

公开(公告)日：2024-02-15

申请号：US18205609

申请日：2023-06-05

Applicant: Google LLC

Inventor： Xinying Song ,  Yang Song

IPC: G06F40/284 ,  G06F16/31 ,  G06F40/40

CPC classification number: G06F40/284 ,  G06F16/322 ,  G06F40/40

Abstract: Systems and methods for performing inference for word or wordpiece tokenization are disclosed using a left-to-right longest-match-first greedy process. In some examples, the vocabulary may be organized into a trie structure in which each node includes a precomputed token or token_ID and a fail link, so that the tokenizer can parse the trie in a single pass to generate a list of only those tokens or token_IDs that correspond to the longest matching vocabulary entries in the sample string, without the need for backtracking. In some examples, the vocabulary may be organized into a trie in which each node has a fail link, and any node that would share token(s) or token_ID(s) of a preceding node is instead given a prev_match link that points back to a chain of nodes with those token(s) or token_ID(s).

公开(公告)号：US20250045534A1

公开(公告)日：2025-02-06

申请号：US18378434

申请日：2023-10-10

Applicant: GOOGLE LLC

Inventor： Swaroop Mishra ,  Ragha Kotikalapudi ,  Sahitya Potluri ,  Taylor Bos ,  YaGuang Li ,  Hanzhao Lin ,  Steven Zheng ,  Yu Du ,  Chen Zhu ,  Chenkai Kuang ,  Xinying Song ,  Heng-Tze Cheng ,  Ed H. Chi ,  Quoc Le

IPC: G06F40/40

Abstract: Implementations relate to a method implemented by one or more processors, the method including: receiving natural language (NL) based input associated with a client device; generating, using a large language model (LLM) and based on processing the NL based input, LLM output; determining, based on the LLM output, a sequence of LLM responses, the sequence of LLM responses including at least one intermediate LLM response and a final LLM response. In some implementations, the method may further include causing the final LLM response to be rendered at the client device. In additional or alternative implementations, the method may further include storing, as an instance of training data for fine-tuning the LLM or an additional LLM, the NL based input along with the final LLM response.

公开(公告)号：US20240362093A1

公开(公告)日：2024-10-31

申请号：US18231606

申请日：2023-08-08

Applicant: GOOGLE LLC

Inventor： Hao Zhou ,  Jamie Hall ,  Xinying Song ,  Sahitya Potluri ,  Yu Du ,  Heng-Tze Cheng ,  Quoc Le ,  Ed H. Chi

IPC: G06F9/54 ,  G06F16/242

CPC classification number: G06F9/547 ,  G06F16/243

Abstract: At least utilizing a custom corpus of documents to condition a large language model (LLM) when generating a response to a user query. In some implementations, a user query associated with a client device is received. An API query for an external application is generated by an LLM based on the user query. The external application has access to a custom corpus of documents comprising a plurality of documents. The external application is queried using the API query. Data representative of one or more documents in the custom corpus of documents is received from the external application in response to the API query. The LLM generates a response to the query that is conditioned on the data representing one or more of the documents in the custom corpus of documents received from the external application. The response to the user query is caused to be rendered on the client device.