INDEX

Explanations

phrases related to misinformation, deception, and false information

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

hens

-0.75

anse

-0.74

foreseen

-0.72

arya

-0.68

area

-0.68

aldo

-0.67

foundation

-0.67

 guiActiveUnfocused

-0.67

illes

-0.67

winner

-0.67

POSITIVE LOGITS

ument

1.05

ulent

0.96

ulence

0.90

 falsely

0.88

 excuse

0.87

 excuses

0.84

 pretext

0.82

 concoct

0.77

pas

0.75

 false

0.74

Activations Density 1.225%

phrases related to misinformation, deception, and false information

No Comments

No Known Activations