INDEX

Explanations

formal expressions or phrases that indicate speaking or discussing a topic

oai_token-act-pair · gpt-3.5-turbo

phrases that reference a way of discussing or presenting information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

-0.76

ufact

-0.75

pload

-0.69

uminati

-0.67

arsity

-0.66

ocene

-0.66

ĸļ

-0.65

icer

-0.64

rium

-0.63

kefeller

-0.62

POSITIVE LOGITS

speaking

0.64

forth

0.64

 though

0.63

 è£ıè

0.62

 however

0.61

Recent

0.60

 Â¯

0.57

IAS

0.57

 suffice

0.57

 excludes

0.57

Activations Density 0.044%

formal expressions or phrases that indicate speaking or discussing a topic

phrases that reference a way of discussing or presenting information

No Comments

No Known Activations