INDEX

Explanations

references to familial relationships, particularly those involving sisters

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

veyard

-0.72

ustomed

-0.68

uden

-0.66

ustom

-0.63

ocobo

-0.62

raltar

-0.61

ered

-0.61

OVER

-0.60

ondon

-0.60

CAST

-0.60

POSITIVE LOGITS

hood

1.22

 sister

0.98

 sisters

0.97

hips

0.93

folk

0.90

 Isabel

0.82

 Kate

0.82

 Carol

0.81

 daughters

0.81

Maria

0.81

Activations Density 0.007%

references to familial relationships, particularly those involving sisters

No Comments

No Known Activations