INDEX

Explanations

mentions of news reports or journalistic activities

oai_token-act-pair · gpt-3.5-turbo

fragments of news reports related to incidents, witnesses, and authorities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Snape

-0.71

 rever

-0.71

 Â¥

-0.70

utor

-0.69

 Malfoy

-0.67

 Rolls

-0.67

igraph

-0.66

sing

-0.66

rower

-0.66

 quitting

-0.65

POSITIVE LOGITS

CBS

1.07

 NEWS

0.96

NEWS

0.94

 NETWORK

0.94

FOX

0.93

 Investigative

0.92

News

0.92

 VIDEOS

0.91

abc

0.91

ABC

0.90

Activations Density 0.320%

mentions of news reports or journalistic activities

fragments of news reports related to incidents, witnesses, and authorities

No Comments

No Known Activations