INDEX

Explanations

superlative expressions indicating extreme experiences or opinions

oai_token-act-pair · gpt-3.5-turbo

strong evaluations and assessments of experiences

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

regor

-0.63

 preferred

-0.61

 resid

-0.55

kas

-0.55

 occasional

-0.54

 Parables

-0.54

 constant

-0.53

Morning

-0.52

 millenn

-0.52

 vital

-0.51

POSITIVE LOGITS

 anywhere

1.13

 EVER

1.06

 imaginable

0.88

 ever

0.84

 thus

0.83

!.

0.81

ever

0.81

!!!!

0.75

istani

0.75

!!

0.75

Activations Density 0.141%

superlative expressions indicating extreme experiences or opinions

strong evaluations and assessments of experiences

No Comments

No Known Activations