INDEX

Explanations

phrases involving people's names and their professional titles or roles

oai_token-act-pair · gpt-3.5-turbo

statements made by various individuals, particularly in a formal or authoritative context

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

 lasts

-0.75

rats

-0.74

 exists

-0.67

 larg

-0.65

commit

-0.64

 calendar

-0.63

 carc

-0.63

 excludes

-0.60

 prev

-0.60

 survives

-0.60

POSITIVE LOGITS

Jr

0.84

 Managing

0.80

 managing

0.78

MD

0.76

 adding

0.75

IPS

0.75

 dean

0.74

Sr

0.74

anca

0.73

İĭ

0.70

Activations Density 0.158%

phrases involving people's names and their professional titles or roles

statements made by various individuals, particularly in a formal or authoritative context

No Comments

No Known Activations