INDEX

Explanations

interviews

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ILCS

-0.93

years

-0.71

 sooner

-0.68

acci

-0.68

 realised

-0.67

angular

-0.65

nav

-0.64

////////////////////////////////

-0.63

reci

-0.63

atin

-0.62

POSITIVE LOGITS

 reporters

1.04

 CNBC

1.00

 HuffPost

0.97

 Wired

0.96

Ars

0.95

 Collider

0.94

 Newsweek

0.94

racuse

0.91

 Rolling

0.91

 Vanity

0.89

Activations Density 0.099%

interviews

No Comments

No Known Activations