INDEX

Explanations

mentions of political debates

oai_token-act-pair · gpt-3.5-turbo

references to political debates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

reated

-0.78

ridges

-0.68

uras

-0.68

icidal

-0.67

relative

-0.66

reads

-0.66

ocument

-0.63

ruce

-0.63

inness

-0.63

YE

-0.62

POSITIVE LOGITS

halla

0.91

 moderators

0.85

 transcripts

0.82

 venue

0.79

 transcript

0.78

nas

0.77

 prep

0.75

 contestant

0.74

 airs

0.72

 Fallon

0.71

Activations Density 0.031%

mentions of political debates

references to political debates

No Comments

No Known Activations