INDEX

Explanations

words related to persistent actions or events

oai_token-act-pair · gpt-3.5-turbo

instances of the verb "stop" and its variations, indicating ongoing actions or the lack of cessation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Skydragon

-0.72

soDeliveryDate

-0.66

ighth

-0.64

 accessed

-0.61

 Presbyterian

-0.60

 advisory

-0.59

lore

-0.59

 Andersen

-0.57

 dated

-0.57

 Mandatory

-0.56

POSITIVE LOGITS

 anymore

0.92

nor

0.83

Bright

0.73

ggle

0.70

natureconservancy

0.65

 elbows

0.65

oshenko

0.64

rays

0.64

steen

0.64

 Canaver

0.63

Activations Density 0.132%

words related to persistent actions or events

instances of the verb "stop" and its variations, indicating ongoing actions or the lack of cessation

No Comments

No Known Activations