INDEX

Explanations

time durations specified in hours

oai_token-act-pair · gpt-3.5-turbo

references to "an hour and a half" or variations thereof

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

antha

-0.68

 Surviv

-0.67

ologically

-0.67

ãĥ¤

-0.65

eus

-0.64

å£«

-0.64

ocrats

-0.63

thumbnails

-0.63

 Prev

-0.63

Reviewer

-0.62

POSITIVE LOGITS

 dozen

0.79

entimes

0.71

hearted

0.69

 glance

0.68

 million

0.67

heartedly

0.67

quel

0.65

 sized

0.64

 gallon

0.63

 overs

0.63

Activations Density 0.016%

time durations specified in hours

references to "an hour and a half" or variations thereof

No Comments

No Known Activations