INDEX

Explanations

proper nouns referencing individuals and organizations

oai_token-act-pair · gpt-3.5-turbo

phrases indicating reported speech or quotations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 favor

-0.82

 favors

-0.78

oros

-0.77

 neighboring

-0.70

imize

-0.70

 favored

-0.69

 favorably

-0.69

colored

-0.68

 gotten

-0.67

 Whitman

-0.66

POSITIVE LOGITS

BBC

1.14

BBC

1.07

 emphas

1.03

 organisers

1.02

 analys

1.01

NRL

1.01

 programme

1.00

 organisation

1.00

 organised

1.00

 recognise

0.97

Activations Density 0.153%

proper nouns referencing individuals and organizations

phrases indicating reported speech or quotations

No Comments

No Known Activations