INDEX

Explanations

continuations in a document or article

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the continuation of an article or story segment

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Ly

-0.81

OH

-0.71

yll

-0.71

 honor

-0.69

aceae

-0.67

rams

-0.67

ops

-0.66

houses

-0.66

mos

-0.66

 happiest

-0.65

POSITIVE LOGITS

 Continued

1.61

ĸļ

0.91

 Transcript

0.90

 Below

0.85

 subsequ

0.84

 unlaw

0.82

 Consent

0.79

 VIDEOS

0.77

theless

0.76

 Parenthood

0.74

Activations Density 0.004%

continuations in a document or article

phrases indicating the continuation of an article or story segment

No Comments

No Known Activations