INDEX

Explanations

specific mentions of stories or news articles

oai_token-act-pair · gpt-3.5-turbo

frequent references to "story" and related narrative elements

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

ignt

-0.89

emale

-0.87

inking

-0.86

ategory

-0.86

ateurs

-0.84

omination

-0.84

amily

-0.83

omin

-0.82

orneys

-0.80

ojure

-0.79

POSITIVE LOGITS

telling

1.06

arc

0.95

 revolving

0.94

 arcs

0.85

board

0.84

boards

0.83

book

0.83

tell

0.79

books

0.79

 Stories

0.78

Activations Density 0.044%

specific mentions of stories or news articles

frequent references to "story" and related narrative elements

No Comments

No Known Activations