INDEX

Explanations

dates and time-related words

oai_token-act-pair · gpt-3.5-turbo

specific titles and names related to media and entertainment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ansas

-0.74

lly

-0.73

aeda

-0.73

 Abedin

-0.70

reth

-0.69

ichita

-0.67

ds

-0.67

 stocking

-0.66

Yad

-0.65

ozo

-0.65

POSITIVE LOGITS

 VIDEOS

1.05

EST

0.92

ICAN

0.91

MY

0.89

LAT

0.85

TER

0.83

TE

0.82

DOWN

0.80

IK

0.80

itudes

0.79

Activations Density 0.017%

dates and time-related words

specific titles and names related to media and entertainment

No Comments

No Known Activations