INDEX

Explanations

comparative phrases indicating something is of a greater degree than what is initially suggested or expected

oai_token-act-pair · gpt-3.5-turbo

references to exceeding expectations or quantities

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

xtap

-0.75

 Nile

-0.69

 Attend

-0.66

 Period

-0.66

uca

-0.65

Rus

-0.63

pex

-0.63

 Translation

-0.62

avis

-0.62

 Maze

-0.61

POSITIVE LOGITS

 than

1.20

 complicated

0.87

 stringent

0.87

 importantly

0.86

 sophisticated

0.85

 manageable

0.85

 refined

0.85

 expensive

0.85

than

0.80

 nuanced

0.77

Activations Density 0.068%

comparative phrases indicating something is of a greater degree than what is initially suggested or expected

references to exceeding expectations or quantities

No Comments

No Known Activations