INDEX

Explanations

numerical values indicating percentage increases or decreases

oai_token-act-pair · gpt-3.5-turbo

numerical statistics or significant percentages related to various metrics

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

sels

-0.64

 Rica

-0.62

 Machines

-0.62

 Leilan

-0.60

ells

-0.59

 fame

-0.59

 Muse

-0.59

yles

-0.58

 lobb

-0.57

 antagonists

-0.55

POSITIVE LOGITS

%-

1.12

0.88

taboola

0.87

th

0.86

0.79

 dimensional

0.78

acious

0.77

 degree

0.76

bedroom

0.76

 percent

0.76

Activations Density 0.123%

numerical values indicating percentage increases or decreases

numerical statistics or significant percentages related to various metrics

No Comments

No Known Activations