INDEX

Explanations

instances of things happening or being chosen without a specific pattern or reason

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "randomly" and related terms about random selection or action

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ties

-0.79

Operation

-0.74

ger

-0.74

 Millennium

-0.73

gers

-0.71

 Agenda

-0.67

iens

-0.67

soc

-0.66

 intentions

-0.66

 Killer

-0.66

POSITIVE LOGITS

 detonated

0.95

 planted

0.92

 randomly

0.89

 sampled

0.88

 combust

0.88

 implanted

0.86

 configured

0.85

 sacrificed

0.85

 chose

0.83

 regenerate

0.82

Activations Density 0.017%

instances of things happening or being chosen without a specific pattern or reason

occurrences of the word "randomly" and related terms about random selection or action

No Comments

No Known Activations