INDEX

Explanations

dates or time-related terms

oai_token-act-pair · gpt-3.5-turbo

specific numerical references, particularly related to years

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

TPP

-0.67

 Loans

-0.65

 Syndicate

-0.64

 Collider

-0.61

DRM

-0.61

obal

-0.60

 Redemption

-0.60

ULE

-0.59

henko

-0.59

ovie

-0.58

POSITIVE LOGITS

Eight

1.11

eteen

1.06

 thousand

1.04

eight

0.98

irteen

0.96

teen

0.94

eenth

0.90

 Thousand

0.89

seven

0.88

 Shades

0.88

Activations Density 0.023%

dates or time-related terms

specific numerical references, particularly related to years

No Comments

No Known Activations