INDEX

Explanations

proper nouns, possibly related to locations and technology

oai_token-act-pair · gpt-3.5-turbo

nouns and terms related to various entities and structures, suggesting a focus on specific items, roles, and categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 disadvant

-0.51

]+

-0.50

farious

-0.49

 freeing

-0.46

$.

-0.45

fitting

-0.45

ãģĻ

-0.45

 Kardash

-0.44

 Levant

-0.42

necessary

-0.42

POSITIVE LOGITS

 consisted

1.08

 consists

1.06

 comprises

0.93

has

0.88

is

0.86

 appeared

0.86

 appears

0.86

may

0.85

 grew

0.84

 cannot

0.84

Activations Density 0.996%

proper nouns, possibly related to locations and technology

nouns and terms related to various entities and structures, suggesting a focus on specific items, roles, and categories

No Comments

No Known Activations