INDEX

Explanations

dates or time-related information

oai_token-act-pair · gpt-3.5-turbo

commas in sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Russ

-0.69

Interested

-0.65

Rew

-0.62

oir

-0.62

Amount

-0.61

iliar

-0.61

OUGH

-0.59

owitz

-0.59

worldly

-0.58

UF

-0.57

POSITIVE LOGITS

 joins

0.89

 withdrew

0.88

 arrives

0.80

 stands

0.77

 greets

0.76

 welcomes

0.76

 owes

0.76

 sits

0.75

 remembers

0.74

 became

0.73

Activations Density 0.260%

dates or time-related information

commas in sentences

No Comments

No Known Activations