INDEX

Explanations

words and phrases related to assessments or judgments, often expressing strong opinions or definitions

oai_token-act-pair · gpt-3.5-turbo

expressions of subjective opinions or evaluations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

cair

-0.66

anmar

-0.63

ieties

-0.63

 Citiz

-0.62

ojure

-0.61

uador

-0.60

lvl

-0.60

vana

-0.59

arnaev

-0.59

kefeller

-0.59

POSITIVE LOGITS

0.74

 unsu

0.73

 unavailable

0.72

 preferable

0.69

 excluded

0.69

 indistinguishable

0.68

 identical

0.67

 incompatible

0.66

 superior

0.66

 obliged

0.65

Activations Density 0.133%

words and phrases related to assessments or judgments, often expressing strong opinions or definitions

expressions of subjective opinions or evaluations

No Comments

No Known Activations