INDEX

Explanations

adjectives with positive connotations placed before the word "but," suggesting a contrast or unexpected element

oai_token-act-pair · gpt-3.5-turbo

the word "but" used in contrasting statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ento

-0.90

oire

-0.87

ORPG

-0.67

eatures

-0.67

itness

-0.66

asions

-0.63

OLOGY

-0.63

utonium

-0.62

axter

-0.62

ogens

-0.62

POSITIVE LOGITS

 nonetheless

1.16

 nevertheless

1.09

tons

1.08

chery

0.99

 still

0.91

 unmist

0.89

 manageable

0.89

 effective

0.88

still

0.86

 undeniably

0.85

Activations Density 0.088%

adjectives with positive connotations placed before the word "but," suggesting a contrast or unexpected element

the word "but" used in contrasting statements

No Comments

No Known Activations