INDEX

Explanations

names of people and titles of positions related to law, military, policy, and expertise

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 stret

-0.80

tumblr

-0.71

 screws

-0.69

MSN

-0.68

nig

-0.68

 matured

-0.66

 toilet

-0.66

 bathroom

-0.66

 transgress

-0.66

abiding

-0.65

POSITIVE LOGITS

earcher

0.94

 sociology

0.91

 consultancy

0.88

 Consulting

0.87

heastern

0.86

 analyst

0.86

research

0.84

 Research

0.84

 researcher

0.83

 specializing

0.81

Activations Density 6.385%

names of people and titles of positions related to law, military, policy, and expertise

No Comments

No Known Activations