INDEX

Explanations

past tense verbs ending in 'ed'

oai_token-act-pair · gpt-3.5-turbo

instances of the word "done" followed by various contexts of completion or finality

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

lar

-0.76

ulates

-0.74

olson

-0.72

liner

-0.71

imov

-0.71

dayName

-0.71

Reviewer

-0.70

ussen

-0.69

lights

-0.69

uyomi

-0.68

POSITIVE LOGITS

pez

1.04

 nothing

0.74

 away

0.71

 differently

0.67

 prepar

0.66

 done

0.65

 condu

0.65

ggie

0.63

 administr

0.63

xxxxxxxx

0.63

Activations Density 0.026%

past tense verbs ending in 'ed'

instances of the word "done" followed by various contexts of completion or finality

No Comments

No Known Activations