INDEX

Explanations

locations and details related to specific regions, incidents, and names of places and people

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

 Reincarn

-0.78

 smoot

-0.70

 crank

-0.69

 positives

-0.68

 apolog

-0.66

 acceler

-0.64

 WATCHED

-0.64

 crew

-0.64

 probes

-0.63

 sentence

-0.63

POSITIVE LOGITS

opolis

1.14

town

1.12

esville

1.08

ville

1.06

 Springs

1.05

burgh

1.05

 Township

1.02

stown

1.00

borough

0.99

 Heights

0.99

Activations Density 1.422%

locations and details related to specific regions, incidents, and names of places and people

No Comments

No Known Activations