INDEX

Explanations

words related to achieving, attaining or reaching

oai_token-act-pair · gpt-3.5-turbo

terms related to acknowledgment or acceptance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

VID

-0.95

prototype

-0.91

ä

-0.81

vous

-0.77

çīĪ

-0.75

trap

-0.74

drm

-0.74

wich

-0.74

éĹĺ

-0.72

 Dare

-0.72

POSITIVE LOGITS

acc

1.24

reditation

1.08

redited

0.94

ursed

0.93

 mathemat

0.93

enture

0.88

ommod

0.86

imate

0.86

acia

0.84

licted

0.81

Activations Density 0.006%

words related to achieving, attaining or reaching

terms related to acknowledgment or acceptance

No Comments

No Known Activations