INDEX

Explanations

words related to cryptography or encryption schemes

oai_token-act-pair · gpt-3.5-turbo

mentions of "ep" and its variations, likely related to episodic content

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.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

Nig

-0.78

 Bowling

-0.68

vette

-0.66

\\\\\\\\\\\\\\\\

-0.66

 Viol

-0.65

jar

-0.65

STON

-0.65

 Gardner

-0.64

 Violent

-0.64

ali

-0.63

POSITIVE LOGITS

ep

3.70

Ep

2.07

Ep

1.88

EP

1.43

ec

1.22

ep

1.19

 epist

1.16

ecc

1.13

 epis

1.10

 epoch

1.09

Activations Density 0.012%

words related to cryptography or encryption schemes

mentions of "ep" and its variations, likely related to episodic content

No Comments

No Known Activations