INDEX

Explanations

words related to working together with others, specifically emphasizing the term "colleagues"

oai_token-act-pair · gpt-3.5-turbo

references to colleagues and teamwork

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

ball

-0.78

aval

-0.76

 Ukrain

-0.71

direction

-0.69

sole

-0.68

forest

-0.67

oted

-0.67

ibaba

-0.67

si

-0.65

uder

-0.65

POSITIVE LOGITS

 colleagues

1.23

 colleague

1.12

 coworkers

0.95

ratulations

0.81

mates

0.80

 classmates

0.80

 roomm

0.76

 collaborators

0.73

 pals

0.70

ority

0.70

Activations Density 0.013%

words related to working together with others, specifically emphasizing the term "colleagues"

references to colleagues and teamwork

No Comments

No Known Activations