INDEX

Explanations

words and phrases that indicate questioning or uncertainty

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ECD

-0.79

cit

-0.77

ande

-0.76

anc

-0.74

venants

-0.69

ABE

-0.66

Nev

-0.65

 Cohn

-0.65

ocally

-0.64

agues

-0.63

POSITIVE LOGITS

 whats

1.27

how

1.11

 what

1.11

 WHAT

1.10

what

1.07

 Whats

0.97

HOW

0.97

how

0.92

WHY

0.92

WHAT

0.90

Activations Density 0.035%

words and phrases that indicate questioning or uncertainty

No Comments

No Known Activations