INDEX

Explanations

words related to conflict or manipulation

oai_token-act-pair · gpt-3.5-turbo

variations of the word "angle."

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

eele

-0.74

grades

-0.71

ibel

-0.71

alty

-0.70

ledged

-0.70

grading

-0.69

units

-0.68

essee

-0.68

uum

-0.67

oji

-0.67

POSITIVE LOGITS

weed

0.94

angle

0.83

 vines

0.80

 carrot

0.78

 sling

0.76

 dangling

0.75

 rope

0.71

 snatch

0.70

weight

0.70

luaj

0.69

Activations Density 0.033%

words related to conflict or manipulation

variations of the word "angle."

No Comments

No Known Activations