INDEX

Explanations

mention of "Hot" in various contexts

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

theless

-0.88

 acknow

-0.80

 conclud

-0.76

ufact

-0.74

ĸļ

-0.72

ajor

-0.71

xual

-0.71

Downloadha

-0.69

 guiActiveUn

-0.69

 confir

-0.69

POSITIVE LOGITS

Spot

1.07

ness

1.05

shots

0.96

dog

0.96

fixes

0.95

fix

0.94

dogs

0.93

keys

0.93

bed

0.90

water

0.90

Activations Density 0.009%

mention of "Hot" in various contexts

No Comments

No Known Activations