INDEX

Explanations

lines containing a specific formatting or encoding pattern

oai_token-act-pair · gpt-3.5-turbo

various forms of numerical values or counts in context

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

hement

-0.64

 Hurt

-0.64

Lit

-0.63

etta

-0.62

iannopoulos

-0.62

utics

-0.61

 Reincarnated

-0.61

avin

-0.61

 mistaken

-0.59

 estranged

-0.58

POSITIVE LOGITS

ÂłÂłÂłÂł

1.16

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

1.15

ÂłÂłÂłÂłÂłÂłÂłÂł

1.05

ÂłÂłÂł

1.01

posted

0.99

ÂłÂł

0.95

Posted

0.84

↵Âł

0.78

TABLE

0.76

Trivia

0.71

Activations Density 0.700%

lines containing a specific formatting or encoding pattern

various forms of numerical values or counts in context

No Comments

No Known Activations