INDEX

Explanations

words related to positive actions or attributes, such as supporting, enhancing, providing incentives, and improving

oai_token-act-pair · gpt-3.5-turbo

positive actions or outcomes related to support and improvement

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

tick

-0.78

tree

-0.72

 helicop

-0.70

oan

-0.69

noon

-0.67

->

-0.66

tu

-0.66

Oo

-0.65

lua

-0.65

fuck

-0.65

POSITIVE LOGITS

 them

0.79

 consequ

0.75

 alike

0.74

soDeliveryDate

0.71

 wider

0.69

 opportunities

0.68

 communities

0.68

 morale

0.67

 consequential

0.67

 broader

0.66

Activations Density 0.325%

words related to positive actions or attributes, such as supporting, enhancing, providing incentives, and improving

positive actions or outcomes related to support and improvement

No Comments

No Known Activations