INDEX

Explanations

words related to time or sequence

oai_token-act-pair · gpt-3.5-turbo

words related to predictions or forecasts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

dye

-0.69

fax

-0.67

Spider

-0.66

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.63

 Deal

-0.62

oshenko

-0.61

witch

-0.60

mens

-0.59

 darn

-0.59

udi

-0.59

POSITIVE LOGITS

itism

0.74

ileged

0.74

ilial

0.73

ency

0.71

SHIP

0.71

ctive

0.69

erate

0.68

ELY

0.68

tenance

0.68

ential

0.67

Activations Density 0.086%

words related to time or sequence

words related to predictions or forecasts

No Comments

No Known Activations