INDEX

Explanations

phrases related to unexpected occurrences or issues that arise

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "popping up."

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

reat

-0.68

aving

-0.68

ft

-0.67

anus

-0.65

WA

-0.65

uckland

-0.62

aped

-0.61

iles

-0.61

gow

-0.60

reet

-0.60

POSITIVE LOGITS

 smoothly

0.73

âĶĢ

0.72

river

0.68

burst

0.68

 nicely

0.68

IER

0.67

 GOODMAN

0.66

 uneven

0.66

EVA

0.65

Ñĭ

0.65

Activations Density 0.130%

phrases related to unexpected occurrences or issues that arise

instances of the phrase "popping up."

No Comments

No Known Activations