INDEX

Explanations

phrases related to taking action or making a change in response to a situation

oai_token-act-pair · gpt-3.5-turbo

phrases suggesting action or concern regarding an issue

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

Constructed

-0.71

 Died

-0.70

 Writ

-0.64

emies

-0.64

omet

-0.64

Creat

-0.64

told

-0.64

 confessed

-0.63

authored

-0.62

iets

-0.62

POSITIVE LOGITS

erous

0.83

 administr

0.75

isance

0.72

arios

0.71

 redist

0.69

 behalf

0.68

 halfway

0.68

 improving

0.65

ILLE

0.65

 injustice

0.65

Activations Density 0.049%

phrases related to taking action or making a change in response to a situation

phrases suggesting action or concern regarding an issue

No Comments

No Known Activations