INDEX

Explanations

phrases related to monitoring or paying attention to specific things or people

oai_token-act-pair · gpt-3.5-turbo

references to monitoring or observing closely

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Whites

-0.74

 Gibbs

-0.70

 substituted

-0.63

 collapsing

-0.62

split

-0.60

 whites

-0.58

channelAvailability

-0.58

 unfocusedRange

-0.56

 Giuliani

-0.56

ania

-0.55

POSITIVE LOGITS

eful

0.82

ocket

0.74

tower

0.74

agascar

0.72

ENSE

0.70

ock

0.70

OSH

0.69

LOCK

0.69

agos

0.69

gency

0.68

Activations Density 0.023%

phrases related to monitoring or paying attention to specific things or people

references to monitoring or observing closely

No Comments

No Known Activations