INDEX

Explanations

words describing positive or negative experiences and emotions

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

aucus

-0.69

arians

-0.63

aan

-0.62

arius

-0.61

inition

-0.61

 Feder

-0.61

ARS

-0.60

govtrack

-0.59

ARE

-0.59

FIL

-0.59

POSITIVE LOGITS

ries

1.29

 surprises

1.04

ness

1.01

 smelling

1.00

lihood

0.98

ties

0.96

 pleasant

0.90

nesses

0.89

ments

0.85

 surpr

0.81

Activations Density 0.013%

words describing positive or negative experiences and emotions

No Comments

No Known Activations