INDEX

Explanations

phrases related to future events or outcomes, especially those that are unlikely or surprising

oai_token-act-pair · gpt-3.5-turbo

occurrences and mentions of the word "happen."

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

edged

-0.74

ondo

-0.74

ilts

-0.73

rador

-0.72

stra

-0.71

otto

-0.70

rete

-0.70

rosse

-0.68

cius

-0.68

acked

-0.67

POSITIVE LOGITS

uate

0.91

rences

0.73

 anytime

0.73

 undone

0.71

Kyl

0.70

uates

0.68

uating

0.68

 happen

0.67

uation

0.67

 anywhere

0.66

Activations Density 0.034%

phrases related to future events or outcomes, especially those that are unlikely or surprising

occurrences and mentions of the word "happen."

No Comments

No Known Activations