INDEX

Explanations

negative statements and references to injuries or casualties

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

erity

-0.84

uble

-0.77

gat

-0.75

Tes

-0.74

DragonMagazine

-0.73

perfect

-0.70

rypt

-0.70

redit

-0.69

otton

-0.69

ptin

-0.69

POSITIVE LOGITS

 indication

1.12

 indications

1.06

 formal

0.92

 conclusive

0.90

 mention

0.89

 injuries

0.87

 kidding

0.87

 official

0.87

 signs

0.87

 immediate

0.86

Activations Density 0.071%

negative statements and references to injuries or casualties

No Comments

No Known Activations