INDEX

Explanations

phrases related to informing or recounting events

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

elsh

-0.74

mun

-0.70

 sidx

-0.70

imposed

-0.70

adesh

-0.68

 ILCS

-0.68

Interstitial

-0.67

nam

-0.65

mite

-0.65

aband

-0.61

POSITIVE LOGITS

us

1.29

tale

1.18

me

1.15

him

1.07

ingly

1.03

 reporters

0.91

 them

0.90

 listeners

0.88

 viewers

0.86

 passers

0.82

Activations Density 2.982%

phrases related to informing or recounting events

No Comments

No Known Activations