INDEX

Explanations

mentions of the year 2020 in a political context

oai_token-act-pair · gpt-3.5-turbo

references to the year 2020

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

iqu

-0.79

ype

-0.72

iera

-0.70

kell

-0.70

oaded

-0.69

hen

-0.68

imen

-0.68

ordes

-0.67

aws

-0.66

POSITIVE LOGITS

0.91

 onwards

0.90

 census

0.85

0.82

0.81

 onward

0.78

0.74

 timeframe

0.72

0.71

0.70

Activations Density 0.010%

mentions of the year 2020 in a political context

references to the year 2020

No Comments

No Known Activations