INDEX

Explanations

references to the year 2001

oai_token-act-pair · gpt-3.5-turbo

references to the year 2001

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

accompan

-0.75

Nation

-0.74

ony

-0.69

atted

-0.69

attled

-0.67

shaw

-0.66

rom

-0.66

agues

-0.66

asts

-0.64

alling

-0.63

POSITIVE LOGITS

ĸļ

0.98

ãĥĥãĥī

0.89

-'

0.84

ãĥ³ãĤ¸

0.78

sov

0.73

etics

0.73

ocide

0.72

aru

0.68

ÙĨ

0.68

ãĥ¼ãĥĨãĤ£

0.67

Activations Density 0.012%

references to the year 2001

references to the year 2001

No Comments

No Known Activations