INDEX

Explanations

actions related to authority figures and directives

oai_token-act-pair · gpt-3.5-turbo

actions related to sending, inviting, and communicating

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

til

-0.74

abiding

-0.72

yond

-0.67

selves

-0.67

inav

-0.66

stellar

-0.66

 selves

-0.66

ateral

-0.64

 âĢº

-0.64

zbollah

-0.63

POSITIVE LOGITS

 himself

1.27

him

1.18

 herself

1.02

his

1.00

her

0.82

 angrily

0.82

 reprim

0.79

him

0.78

me

0.78

 admon

0.77

Activations Density 0.533%

actions related to authority figures and directives

actions related to sending, inviting, and communicating

No Comments

No Known Activations