INDEX

Explanations

names of political figures

oai_token-act-pair · gpt-3.5-turbo

references to individuals and their affiliations

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

itiveness

-0.58

 DRAGON

-0.55

 hereafter

-0.54

rontal

-0.54

 blackout

-0.54

 latch

-0.54

 Sail

-0.54

akura

-0.52

 entitle

-0.51

iliary

-0.51

POSITIVE LOGITS

steen

0.89

 Canaver

0.79

igham

0.74

endi

0.69

ornia

0.69

ãĤ¼ãĤ¦ãĤ¹

0.69

phia

0.67

enhagen

0.67

esi

0.65

uria

0.65

Activations Density 0.397%

names of political figures

references to individuals and their affiliations

No Comments

No Known Activations