INDEX

Explanations

words related to specific names and identifiers, likely in a context of branding or product lines

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

 nationally

-0.81

 spoken

-0.79

 welfare

-0.71

 disadvantaged

-0.71

 needy

-0.69

 future

-0.69

 favorable

-0.68

 predetermined

-0.68

 medically

-0.67

 statewide

-0.67

POSITIVE LOGITS

osaurus

1.07

opter

1.04

ium

0.96

oop

0.94

ote

0.93

ix

0.92

odon

0.91

ox

0.90

VPN

0.89

omatic

0.89

Activations Density 0.095%

words related to specific names and identifiers, likely in a context of branding or product lines

No Comments

No Known Activations