INDEX

Explanations

words related to proper nouns, specifically names

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of a specific name

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

IBLE

-0.87

ãĥ¼ãĥ³

-0.80

ACTED

-0.79

ISC

-0.75

ources

-0.73

Ø§ÙĦ

-0.72

 Gutenberg

-0.70

 Reaper

-0.68

ãĥ¯

-0.68

 FEMA

-0.67

POSITIVE LOGITS

loo

1.02

ky

1.01

ansky

0.86

aky

0.81

kees

0.81

lip

0.81

usha

0.81

pton

0.80

aku

0.79

Ky

0.79

Activations Density 0.015%

words related to proper nouns, specifically names

repeated mentions of a specific name

No Comments

No Known Activations