INDEX

Explanations

references to the name "Jack"

oai_token-act-pair · gpt-3.5-turbo

references to the word "jack" in various contexts, including expressions and phrases

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

 Frag

-0.79

 Corpus

-0.79

 Rite

-0.71

 Apocalypse

-0.69

 Imper

-0.68

ndra

-0.67

 Whole

-0.63

 livest

-0.63

 Codex

-0.63

 Diet

-0.63

POSITIVE LOGITS

pots

1.14

hammer

1.03

knife

0.99

hole

0.99

ument

0.98

pot

0.96

ety

0.95

kins

0.95

wagon

0.94

owitz

0.94

Activations Density 0.015%

references to the name "Jack"

references to the word "jack" in various contexts, including expressions and phrases

No Comments

No Known Activations