INDEX

Explanations

questions addressed to the recipient

oai_token-act-pair · gpt-3.5-turbo

questions and expressions of curiosity involving the pronouns "you" and "we."

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

-0.74

ruciating

-0.69

 Flavoring

-0.67

atlantic

-0.64

 Background

-0.63

 Patterns

-0.63

ãĤ¦ãĤ¹

-0.60

 Squid

-0.59

 Atlantic

-0.59

links

-0.58

POSITIVE LOGITS

 accept

0.82

 dare

0.80

 succeed

0.80

 recognise

0.79

 abandon

0.77

be

0.75

?'

0.75

?'"

0.74

 presume

0.74

 recons

0.74

Activations Density 0.076%

questions addressed to the recipient

questions and expressions of curiosity involving the pronouns "you" and "we."

No Comments

No Known Activations