INDEX

Explanations

words related to leadership and command

oai_token-act-pair · gpt-3.5-turbo

references to the word "Helm" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 recess

-0.77

 infringing

-0.75

irie

-0.68

 Tribe

-0.68

Republic

-0.67

 synd

-0.67

 Verizon

-0.67

 amendments

-0.67

 Viet

-0.66

 creek

-0.66

POSITIVE LOGITS

 Helm

3.92

 helm

2.28

helm

1.98

Hel

1.70

 Helmet

1.67

HEL

1.46

hel

1.36

 helmets

1.35

endor

1.15

Hel

1.12

Activations Density 0.044%

words related to leadership and command

references to the word "Helm" and its variations in different contexts

No Comments

No Known Activations