INDEX

Explanations

phrases related to damage or harm caused, particularly in environmental contexts

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

soDeliveryDate

-0.90

Ü

-0.84

soType

-0.74

 Dating

-0.71

sonian

-0.69

IQ

-0.69

ulhu

-0.69

urnal

-0.68

udicrous

-0.66

approved

-0.66

POSITIVE LOGITS

 inflicted

1.57

 incurred

1.24

 griev

1.14

 wounds

1.09

 suffered

1.08

 harmed

1.07

 harm

1.07

 inflicting

1.03

 injuries

1.02

 hurting

1.01

Activations Density 0.139%

phrases related to damage or harm caused, particularly in environmental contexts

No Comments

No Known Activations