INDEX

Explanations

phrases indicating a potential error or misguided action

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate mistakes or misunderstandings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Reborn

-0.78

ãĥ´ãĤ¡

-0.72

Done

-0.67

HQ

-0.67

accompan

-0.65

hots

-0.63

ongo

-0.63

 Roses

-0.63

Working

-0.62

oken

-0.61

POSITIVE LOGITS

 assume

1.53

 conclude

1.49

 presume

1.47

 equate

1.46

 dismiss

1.41

 overlook

1.39

 underestimate

1.31

 infer

1.30

 confuse

1.29

 speculate

1.26

Activations Density 0.283%

phrases indicating a potential error or misguided action

phrases that indicate mistakes or misunderstandings

No Comments

No Known Activations