INDEX

Explanations

personal pronouns, especially those indicating singularity such as 'one,' 'no one,' and 'nobody.'

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

tnc

-0.77

ult

-0.72

urations

-0.68

iger

-0.67

 Resurrection

-0.67

ulk

-0.64

ISS

-0.59

urrection

-0.59

 Pelicans

-0.58

 Demons

-0.58

POSITIVE LOGITS

 else

2.04

else

1.47

 Else

1.47

Else

1.40

 knows

0.99

who

0.98

 acquainted

0.88

omever

0.87

 remembers

0.86

 cared

0.85

Activations Density 1.164%

personal pronouns, especially those indicating singularity such as 'one,' 'no one,' and 'nobody.'

No Comments

No Known Activations