INDEX

Explanations

phrases related to negative or problematic behavior

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "behaviour" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

kos

-0.71

milo

-0.70

vu

-0.67

inka

-0.67

ixed

-0.67

mand

-0.66

pop

-0.66

sis

-0.65

riel

-0.64

oiler

-0.64

POSITIVE LOGITS

 behavior

1.02

 behavi

0.97

 behaviour

0.94

 behaviors

0.91

 modification

0.90

avior

0.89

uations

0.88

uation

0.87

 behaviours

0.87

 behav

0.85

Activations Density 0.018%

phrases related to negative or problematic behavior

occurrences of the word "behaviour" in various contexts

No Comments

No Known Activations