INDEX

Explanations

words related to facts, factual information, or investigations

oai_token-act-pair · gpt-3.5-turbo

references to factual information and assertions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Skies

-0.69

 Byrne

-0.67

 Carnage

-0.64

 pierced

-0.63

 buffet

-0.62

Kem

-0.62

 anger

-0.61

 blocked

-0.60

Led

-0.60

ed

-0.59

POSITIVE LOGITS

fact

4.59

facts

2.88

Fact

2.37

factor

1.75

 Fact

1.36

truth

1.32

ifact

1.30

 fact

1.21

Factor

1.15

acts

1.07

Activations Density 0.010%

words related to facts, factual information, or investigations

references to factual information and assertions

No Comments

No Known Activations