INDEX

Explanations

years

oai_token-act-pair · gpt-3.5-turbo

references to specific years, particularly 1996 and 1997

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

hop

-0.80

hatt

-0.75

giving

-0.72

tro

-0.72

holder

-0.70

Tu

-0.69

ysc

-0.68

imble

-0.68

hed

-0.67

groups

-0.67

POSITIVE LOGITS

ĸļ

0.81

-'

0.77

uncture

0.71

ãĥĥãĥī

0.68

ãĤ¦ãĤ¹

0.67

å¹

0.67

ãĥ¼ãĥ«

0.66

aneously

0.66

ords

0.66

âķĲ

0.65

Activations Density 0.016%

years

references to specific years, particularly 1996 and 1997

No Comments

No Known Activations