INDEX

Explanations

the word "below" in various contexts

oai_token-act-pair · gpt-3.5-turbo

the repetitive phrase "below" indicating content or references located lower in the document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

imm

-0.73

forced

-0.72

ogging

-0.71

ãĥı

-0.70

mails

-0.69

cycles

-0.67

ãĥĥ

-0.67

olid

-0.66

Lug

-0.66

minist

-0.66

POSITIVE LOGITS

below

0.88

 below

0.88

 screenshot

0.78

 snippet

0.76

mentioned

0.74

 Below

0.73

above

0.71

 paragraph

0.71

 BELOW

0.71

 infographic

0.68

Activations Density 0.047%

the word "below" in various contexts

the repetitive phrase "below" indicating content or references located lower in the document

No Comments

No Known Activations