INDEX

Explanations

adjectives or adverbs that express value judgments

oai_token-act-pair · gpt-3.5-turbo

adjectives and descriptors that convey strong qualifiers or evaluations

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

aminer

-0.62

udeb

-0.59

ioxide

-0.59

endment

-0.56

Revolution

-0.56

iphate

-0.56

ilet

-0.54

umo

-0.53

RPG

-0.53

ourke

-0.51

POSITIVE LOGITS

etheless

0.95

 enough

0.67

entimes

0.65

ortunately

0.65

nesses

0.62

 amounts

0.59

ly

0.59

 ones

0.59

 apologies

0.58

 importantly

0.57

Activations Density 0.620%

adjectives or adverbs that express value judgments

adjectives and descriptors that convey strong qualifiers or evaluations

No Comments

No Known Activations