INDEX

Explanations

conjunctions, particularly the word "and."

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

stocks

-1.02

ciation

-0.72

Contents

-0.71

IGHTS

-0.71

Ãį

-0.71

isks

-0.69

Versions

-0.69

Ô

-0.68

Physical

-0.68

Sov

-0.68

POSITIVE LOGITS

Tim

1.05

 Justin

1.02

 Joshua

1.01

Ben

0.98

 Melvin

0.98

 Kevin

0.98

 Samuel

0.97

Guy

0.97

Ma

0.97

 Michelle

0.96

Activations Density 0.070%

conjunctions, particularly the word "and."

No Comments

No Known Activations