INDEX

Explanations

the repeated use of the word "back."

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

 delegation

-0.70

viz

-0.62

aucus

-0.58

 Disclosure

-0.56

 valuation

-0.53

CEO

-0.53

ISION

-0.53

 Parables

-0.53

querque

-0.53

 NEWS

-0.52

POSITIVE LOGITS

wards

1.25

packs

1.06

ward

1.01

stab

0.96

fired

0.93

fires

0.91

lash

0.90

spin

0.89

gam

0.88

)=(

0.87

Activations Density 0.031%

the repeated use of the word "back."

No Comments

No Known Activations