INDEX

Explanations

words related to suppression or stifling

oai_token-act-pair · gpt-3.5-turbo

words related to oppression or constraints

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ahar

-0.73

à¥

-0.67

VID

-0.65

 Podesta

-0.65

 Guest

-0.65

 alloc

-0.64

Isa

-0.64

 Ballard

-0.63

ileaks

-0.63

 boarded

-0.63

POSITIVE LOGITS

ifle

3.15

ifles

2.57

ifling

2.32

ifled

2.23

 suppress

1.94

 suppressed

1.38

 repe

1.34

 muzzle

1.32

 carb

1.29

 suppressing

1.25

Activations Density 0.037%

words related to suppression or stifling

words related to oppression or constraints

No Comments

No Known Activations