INDEX

Explanations

phrases or terms related to understanding and comprehension

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

onies

-0.82

gob

-0.76

hire

-0.75

ONS

-0.74

unal

-0.71

ovie

-0.68

arious

-0.65

 exting

-0.64

iere

-0.64

-+-+-+-+

-0.64

POSITIVE LOGITS

how

1.18

why

1.11

WHY

1.10

 whats

0.94

why

0.94

 what

0.85

HOW

0.85

 Understanding

0.84

how

0.78

 nuances

0.77

Activations Density 0.041%

phrases or terms related to understanding and comprehension

No Comments

No Known Activations