INDEX

Explanations

specific references to something being the subject of discussion or analysis

oai_token-act-pair · gpt-3.5-turbo

references to a specific topic or concept that is being emphasized

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

amp

-0.72

esm

-0.69

owed

-0.68

adle

-0.68

ortment

-0.66

rollers

-0.66

wives

-0.66

ahs

-0.65

Band

-0.65

 Scores

-0.64

POSITIVE LOGITS

 particular

1.11

 trope

1.01

 phenomenon

1.01

 predicament

0.97

 newfound

0.97

 discrepancy

0.95

 arrangement

0.91

 arrang

0.90

 outcome

0.86

 omission

0.85

Activations Density 0.215%

specific references to something being the subject of discussion or analysis

references to a specific topic or concept that is being emphasized

No Comments

No Known Activations