INDEX

Explanations

adjectives ending in '-ional'

oai_token-act-pair · gpt-3.5-turbo

words that indicate a fact-based or factual nature

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

tower

-0.85

uden

-0.78

Blog

-0.75

conom

-0.72

KO

-0.68

hook

-0.67

================

-0.66

RGB

-0.66

gone

-0.65

ARDS

-0.65

POSITIVE LOGITS

ities

1.15

ism

1.05

ization

0.99

isations

0.97

ized

0.97

ised

0.93

izational

0.92

idad

0.91

isation

0.91

ists

0.90

Activations Density 0.016%

adjectives ending in '-ional'

words that indicate a fact-based or factual nature

No Comments

No Known Activations