INDEX

Explanations

mentions of the word "itter" with varying degrees of emphasis, possibly indicating a preference for higher activations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "itter" in varying 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

lez

-0.73

ournals

-0.68

urat

-0.67

cephal

-0.66

 Kingdoms

-0.66

Wan

-0.59

 Reconstruction

-0.58

Pakistan

-0.58

Ro

-0.58

---------------

-0.57

POSITIVE LOGITS

geist

1.08

cair

0.97

bug

0.92

sburg

0.88

mite

0.83

tainment

0.80

bugs

0.78

iver

0.76

ing

0.76

ite

0.76

Activations Density 0.016%

mentions of the word "itter" with varying degrees of emphasis, possibly indicating a preference for higher activations

occurrences of the word "itter" in varying contexts

No Comments

No Known Activations