2/ Political Biases Found in NLP Models – measures media biases in LLMs, including the fairness of downstream NLP models tuned on top of politically biased LLMs; reveals that LLMs have political leanings which reinforce existing polarization in corpora.
@dair_ai
-
Top ML Papers Week: Agents, LLMs and Industrial Applications
By
–
Top ML Papers of the Week (August 7 – August 13): – AgentBench
– NeuroImagen
– Trustworthy LLMs
– Evaluating LLMs as Agents
– LLMs for Industrial Control
– LLMs as Database Administrators
… -
LLMs as Database Administrators: Maintenance and Diagnosis
By
–
1/ LLMs as DBA – uses LLMs to acquire database maintenance experience from text; performs DB maintenance knowledge detection from documents & tools, tree of thought reasoning for root cause analysis, and collaborative diagnosis using multiple LLMs.
-
PaLM: Understanding One of AI’s Most Advanced Language Models
By
–
PaLM is one of the most advanced LLMs. PaLM achieved remarkable results on a variety of tasks and has inspired other works such as PaLM 2, Med-PaLM, and all sorts of real-world applications, including Bard. There is no doubt about the impact of PaLM so understanding this
-
LLM Self-Check: Zero-Shot Verification for Complex Reasoning
By
–
8/ Self-Check – explores whether LLMs can perform self-checks which is required for complex tasks that depend on non-linear thinking and multi-step reasoning; it proposes a zero-shot verification scheme to recognize errors without external resources.
-
AutoRobotics-Zero Discovers Adaptive Robot Control Policies
By
–
10/ AutoRobotics-Zero – discovers zero-shot adaptable policies from scratch that enable adaptive behaviors necessary for sudden environmental changes; as an example, the authors demonstrate the automatic discovery of Python code for controlling a robot.
-
Agent Learns Multimodal World Model with Language Predictions
By
–
9/ Agents Model the World with Language – presents an agent that learns a multimodal world model that predicts future text and image representations; it learns to predict future language, video, and rewards.
-
OpenFlamingo: Open-Source Vision-Language Models Family Released
By
–
6/ OpenFlamingo – introduces a family of autoregressive vision-language models ranging from 3B to 9B parameters; the technical report describes the models, training data, and evaluation suite.
-
The Hydra Effect: Self-Repairing Properties in Language Models
By
–
7/ The Hydra Effect – shows that language models exhibit self-repairing properties — when one layer of attention heads is ablated it causes another later layer to take over its function.
-
MetaGPT: LLM-Based Multi-Agent Framework Encoding Human SOPs
By
–
5/ MetaGPT – a framework involving LLM-based multi-agents that encodes human standardized operating procedures (SOPs) to extend complex problem-solving capabilities that mimic efficient human workflows.