INDEX

Explanations

phrases indicating a sequence or action that should occur prior to another action

oai_token-act-pair · gpt-3.5-turbo

phrases indicating necessary actions or information to be addressed prior to a specific event

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

aez

-0.72

aren

-0.72

OTOS

-0.70

DIT

-0.69

hack

-0.68

SA

-0.67

Writer

-0.64

MER

-0.64

PROV

-0.63

utter

-0.63

POSITIVE LOGITS

hand

0.81

 realizing

0.80

 committing

0.77

 sunset

0.75

 embark

0.75

 midnight

0.75

 sunrise

0.74

 settling

0.73

 noon

0.73

 proceeding

0.69

Activations Density 0.064%

phrases indicating a sequence or action that should occur prior to another action

phrases indicating necessary actions or information to be addressed prior to a specific event

No Comments

No Known Activations