INDEX

Explanations

relevant nouns

oai_token-act-pair · gpt-3.5-turbo

references to titles, statistics, and numerical data

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Was

-0.84

Was

-0.80

lished

-0.71

expected

-0.71

was

-0.71

 Added

-0.68

Said

-0.67

printed

-0.66

WAS

-0.66

pared

-0.65

POSITIVE LOGITS

are

1.55

 reside

1.44

 comprise

1.34

 occupy

1.34

 belong

1.32

 aren

1.31

 constitute

1.28

 operate

1.24

 resemble

1.24

 rely

1.23

Activations Density 0.615%

relevant nouns

references to titles, statistics, and numerical data

No Comments

No Known Activations