INDEX

Explanations

words related to human behavior

oai_token-act-pair · gpt-3.5-turbo

references to behavioral patterns or conduct

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

ixed

-0.73

avez

-0.72

kos

-0.72

shall

-0.70

chin

-0.70

source

-0.69

plus

-0.69

bring

-0.68

occup

-0.68

forth

-0.67

POSITIVE LOGITS

 behavior

1.35

 behaviors

1.22

 behavi

1.19

 Behavior

1.12

 behaviour

1.11

avior

1.10

 behaviours

1.08

behavior

0.99

 modification

0.97

 behav

0.85

Activations Density 0.012%

words related to human behavior

references to behavioral patterns or conduct

No Comments

No Known Activations