INDEX

Explanations

references to decorative elements or embellishments

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

ly

-0.92

hold

-0.81

leness

-0.81

REP

-0.81

\\\\\\\\

-0.77

zee

-0.76

 Generations

-0.75

Connector

-0.74

LY

-0.74

LINE

-0.73

POSITIVE LOGITS

arial

1.08

anguage

1.06

mith

1.04

ipop

0.91

urgical

0.90

ize

0.90

pine

0.89

urgy

0.89

chemy

0.89

pha

0.87

Activations Density 0.022%

references to decorative elements or embellishments

No Comments

No Known Activations