INDEX

Explanations

threat-related keywords and phrases

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

arist

-0.90

urgy

-0.86

angles

-0.77

coat

-0.75

 Band

-0.75

cise

-0.74

ricks

-0.74

mys

-0.74

ributes

-0.72

alt

-0.71

POSITIVE LOGITS

 posed

0.99

crow

0.97

 threats

0.90

 threat

0.88

threat

0.84

 glare

0.83

lessly

0.82

 Threat

0.81

mong

0.81

 intimidation

0.80

Activations Density 0.747%

threat-related keywords and phrases

No Comments

No Known Activations