INDEX

Explanations

the word "nanny" with different levels of emphasis

oai_token-act-pair · gpt-3.5-turbo

references to the name "Manny" and related terms

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

orescent

-0.81

icter

-0.80

ueless

-0.79

uctor

-0.79

ENGTH

-0.76

uther

-0.73

ACTED

-0.72

ension

-0.71

itude

-0.71

urdue

-0.70

POSITIVE LOGITS

anny

0.92

moon

0.73

cott

0.72

vine

0.67

cia

0.67

Wan

0.66

Pac

0.66

roid

0.66

cock

0.65

illon

0.65

Activations Density 0.018%

the word "nanny" with different levels of emphasis

references to the name "Manny" and related terms

No Comments

No Known Activations