INDEX

Explanations

introductory phrases indicating contrast or opposition

oai_token-act-pair · gpt-3.5-turbo

phrases that introduce contrasting or caveating statements

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

tnc

-0.73

Jew

-0.66

itto

-0.64

enter

-0.64

jew

-0.64

Jew

-0.63

dal

-0.62

taboola

-0.61

ragon

-0.61

lean

-0.60

POSITIVE LOGITS

 acknowledging

0.97

 technically

0.82

ts

0.81

 conced

0.80

 disagreements

0.69

 admittedly

0.68

lihood

0.68

 initially

0.68

 agreeing

0.67

REDACTED

0.66

Activations Density 0.040%

introductory phrases indicating contrast or opposition

phrases that introduce contrasting or caveating statements

No Comments

No Known Activations