INDEX

Explanations

references to public figures making speeches or statements

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ILCS

-0.77

anny

-0.61

involved

-0.60

ardo

-0.59

wcsstore

-0.59

 broom

-0.57

————

-0.57

 equilibrium

-0.57

=~

-0.56

pend

-0.56

POSITIVE LOGITS

 extensively

1.06

 briefly

0.85

 about

0.77

 backstage

0.74

 passionately

0.73

 frankly

0.72

 onstage

0.69

 candid

0.68

 separately

0.67

 harshly

0.66

Activations Density 0.063%

references to public figures making speeches or statements

No Comments

No Known Activations