INDEX

Explanations

news outlets and topics related to various fields such as engineering, entertainment, and sociology

oai_token-act-pair · gpt-3.5-turbo

the acronym "OM" consistently across various contexts

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

wall

-0.78

hs

-0.69

yards

-0.68

issance

-0.66

âĸ¬

-0.66

place

-0.64

 Ventura

-0.62

wb

-0.61

esville

-0.61

ivities

-0.61

POSITIVE LOGITS

EGA

1.07

orrow

0.92

oS

0.92

ESA

0.91

ulus

0.91

obile

0.90

atis

0.87

ISSION

0.86

psons

0.85

icro

0.84

Activations Density 0.016%

news outlets and topics related to various fields such as engineering, entertainment, and sociology

the acronym "OM" consistently across various contexts

No Comments

No Known Activations