INDEX

Explanations

words related to retractions or corrections

oai_token-act-pair · gpt-3.5-turbo

the word "ret" in various contexts, suggesting a focus on terms related to retention or reference

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

INESS

-0.78

tips

-0.70

¬¼

-0.69

SHIP

-0.68

STEM

-0.67

STD

-0.65

latest

-0.65

 Awakens

-0.63

ï¸

-0.63

WAY

-0.62

POSITIVE LOGITS

ainer

0.99

rans

0.95

ribut

0.94

upt

0.92

arations

0.92

ention

0.92

raction

0.91

tell

0.90

reating

0.89

itled

0.88

Activations Density 0.008%

words related to retractions or corrections

the word "ret" in various contexts, suggesting a focus on terms related to retention or reference

No Comments

No Known Activations