INDEX

Explanations

phrases related to updates and new information in a narrative context

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

tein

-0.85

pill

-0.68

 raping

-0.67

sis

-0.63

 virginity

-0.63

llers

-0.62

ĸļ

-0.62

ħĭ

-0.61

 inferior

-0.59

ocker

-0.58

POSITIVE LOGITS

 developments

0.97

 updates

0.89

 confirm

0.80

catentry

0.79

 confir

0.78

 details

0.78

 Updates

0.76

 progresses

0.76

 develops

0.75

 publishes

0.73

Activations Density 0.086%

phrases related to updates and new information in a narrative context

No Comments

No Known Activations