INDEX

Explanations

the word "myself"

oai_token-act-pair · gpt-3.5-turbo

references to self-identity and personal experiences

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ories

-0.79

ulton

-0.78

olid

-0.77

cemic

-0.71

heny

-0.70

grade

-0.70

orie

-0.67

ibaba

-0.65

 Sierra

-0.62

*/(

-0.62

POSITIVE LOGITS

selves

1.02

 myself

0.99

 personally

0.94

self

0.92

 tremend

0.90

 enthusi

0.85

imei

0.80

 ashamed

0.77

 honoured

0.76

 instinct

0.75

Activations Density 0.010%

the word "myself"

references to self-identity and personal experiences

No Comments

No Known Activations