INDEX

Explanations

references to various forms of violence

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

sonian

-0.86

ocular

-0.86

arton

-0.79

dit

-0.75

oplan

-0.75

é¾įå

-0.74

ramer

-0.73

IX

-0.68

odes

-0.68

ITNESS

-0.67

POSITIVE LOGITS

 perpetrated

1.14

 inflicted

0.98

fighting

0.91

 against

0.89

 violence

0.86

 suppression

0.82

 directed

0.81

 retaliation

0.80

 rampage

0.80

 prevention

0.79

Activations Density 0.021%

references to various forms of violence

No Comments

No Known Activations