INDEX

Explanations

words related to political and controversial social topics

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

ĸļ

-0.76

PDATE

-0.74

AUD

-0.67

 crunch

-0.65

STD

-0.65

NPR

-0.64

actionDate

-0.63

âĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪâĸĪ

-0.63

Lear

-0.63

 fortunate

-0.62

POSITIVE LOGITS

phia

1.02

tein

0.89

ensis

0.88

hire

0.86

 Parish

0.84

 Palace

0.84

utics

0.83

Â®

0.81

oulos

0.77

pta

0.76

Activations Density 2.549%

words related to political and controversial social topics

No Comments

No Known Activations