INDEX

Explanations

positive adjectives and descriptors expressing approval or admiration

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

ngth

-0.93

rive

-0.81

ividual

-0.79

lished

-0.70

opez

-0.66

alks

-0.66

perty

-0.65

usalem

-0.65

cipl

-0.64

assemb

-0.64

POSITIVE LOGITS

soType

0.81

 enough

0.80

 considering

0.79

 ðŁĻĤ

0.74

 explan

0.73

ECA

0.72

LY

0.71

 NEWS

0.70

 ðŁĺ

0.69

XD

0.68

Activations Density 0.101%

positive adjectives and descriptors expressing approval or admiration

No Comments

No Known Activations