INDEX

Explanations

instances where someone is being approached or contacted

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "approached"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

pite

-1.10

extra

-0.80

cell

-0.80

cake

-0.72

fac

-0.69

inker

-0.69

hal

-0.68

oil

-0.68

free

-0.67

fo

-0.66

POSITIVE LOGITS

 approached

1.15

IAS

0.76

OHN

0.76

 Approach

0.75

 Ceres

0.74

ISION

0.73

ison

0.71

ODY

0.70

 cautiously

0.70

URES

0.70

Activations Density 0.007%

instances where someone is being approached or contacted

occurrences of the word "approached"

No Comments

No Known Activations