INDEX

Explanations

contractions and negations

oai_token-act-pair · gpt-3.5-turbo

negative contractions and phrases expressing inability or doubt

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

senal

-0.67

ature

-0.66

CI

-0.61

 arsen

-0.59

imity

-0.56

irie

-0.56

IGHT

-0.56

side

-0.56

ILA

-0.55

cycl

-0.55

POSITIVE LOGITS

 possibly

0.96

 conceive

0.95

 afford

0.92

 resist

0.91

 muster

0.90

 foresee

0.89

 withstand

0.84

 conce

0.81

 stomach

0.77

handle

0.76

Activations Density 0.052%

contractions and negations

negative contractions and phrases expressing inability or doubt

No Comments

No Known Activations