INDEX

Explanations

studies or research findings

oai_token-act-pair · gpt-3.5-turbo

phrases indicating that studies have provided evidence or results

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

iche

-0.91

 contrace

-0.70

pour

-0.70

dt

-0.68

orrow

-0.67

aband

-0.66

externalActionCode

-0.65

EStreamFrame

-0.63

coe

-0.61

etheless

-0.60

POSITIVE LOGITS

ered

0.83

 probable

0.77

 favour

0.72

 conclusive

0.71

rays

0.69

-+-+

0.69

runners

0.69

 abnormalities

0.68

awed

0.67

rooms

0.67

Activations Density 0.054%

studies or research findings

phrases indicating that studies have provided evidence or results

No Comments

No Known Activations