INDEX

Explanations

phrases related to political events, polling locations, and governmental actions

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

itives

-0.73

ptions

-0.65

earable

-0.65

 digs

-0.62

SPONSORED

-0.61

 pains

-0.61

Finish

-0.60

MpServer

-0.60

ptive

-0.60

ickets

-0.60

POSITIVE LOGITS

 virtue

1.23

 contrast

1.15

akuya

1.08

products

1.04

catch

1.03

 implication

1.00

product

0.98

stand

0.92

 default

0.91

pass

0.90

Activations Density 0.253%

phrases related to political events, polling locations, and governmental actions

No Comments

No Known Activations