INDEX

Explanations

unusual or peculiar concepts or situations

oai_token-act-pair · gpt-3.5-turbo

terms related to strangeness or abnormality

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

cussion

-0.80

vation

-0.79

bers

-0.78

apers

-0.78

payers

-0.77

bern

-0.77

chen

-0.76

amaru

-0.74

adr

-0.74

payer

-0.73

POSITIVE LOGITS

ly

0.97

 occurrences

0.94

 coincidence

0.80

ness

0.80

 twists

0.78

 anomalies

0.77

 twist

0.74

worldly

0.72

 Ares

0.72

 aber

0.71

Activations Density 0.032%

unusual or peculiar concepts or situations

terms related to strangeness or abnormality

No Comments

No Known Activations