INDEX

Explanations

negative statements or contradictions

oai_token-act-pair · gpt-3.5-turbo

negations or phrases indicating what something is not

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

Continue

-0.68

Untitled

-0.66

cano

-0.66

psc

-0.64

âĹ¼

-0.62

 Tens

-0.61

Vers

-0.61

ounters

-0.60

holder

-0.59

RTX

-0.58

POSITIVE LOGITS

 necessarily

1.32

 merely

1.09

isse

0.91

 infall

0.91

 inherently

0.82

withstanding

0.82

 solely

0.82

 simply

0.78

elia

0.76

icably

0.76

Activations Density 0.301%

negative statements or contradictions

negations or phrases indicating what something is not

No Comments

No Known Activations