INDEX

Explanations

phrases indicating a lack of alignment or correspondence between different entities or perspectives

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reflect" and its variations, indicating a focus on the concept of reflection or representation in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ciating

-0.83

anch

-0.81

kees

-0.72

 externalToEVAOnly

-0.70

queue

-0.69

 Unloaded

-0.68

sites

-0.68

ccess

-0.67

weed

-0.65

efer

-0.65

POSITIVE LOGITS

ively

0.95

 reflected

0.84

 sunlight

0.83

 reflections

0.80

iveness

0.77

orical

0.71

 reflect

0.70

 reflection

0.70

 opin

0.69

iments

0.68

Activations Density 0.034%

phrases indicating a lack of alignment or correspondence between different entities or perspectives

instances of the word "reflect" and its variations, indicating a focus on the concept of reflection or representation in various contexts

No Comments

No Known Activations