INDEX

Explanations

words related to gaining entry or access

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

 Archangel

-0.69

 limb

-0.64

ijn

-0.64

 Kern

-0.62

 shred

-0.59

 estranged

-0.59

pend

-0.58

lication

-0.58

heet

-0.57

 Gael

-0.57

POSITIVE LOGITS

prising

1.02

ways

0.99

prises

0.96

tained

0.96

tain

0.94

door

0.93

tainment

0.92

ibility

0.91

APE

0.91

prise

0.90

Activations Density 2.189%

words related to gaining entry or access

No Comments

No Known Activations