INDEX

Explanations

the word "very" followed by positive or neutral expressions

oai_token-act-pair · gpt-3.5-turbo

expressions of strong emotions or sentiments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

adelphia

-0.77

olor

-0.76

onis

-0.76

sburgh

-0.71

ourses

-0.68

osal

-0.67

zan

-0.67

heid

-0.67

 Cheong

-0.66

ensis

-0.66

POSITIVE LOGITS

 important

0.95

 informative

0.91

 difficult

0.91

 rare

0.88

 interesting

0.88

 exciting

0.87

 useful

0.85

 unlikely

0.85

 handy

0.84

 readable

0.83

Activations Density 0.064%

the word "very" followed by positive or neutral expressions

expressions of strong emotions or sentiments

No Comments

No Known Activations