INDEX

Explanations

mentions of companies and organizations with a strong positive or negative association

oai_token-act-pair · gpt-3.5-turbo

sentences that include a high frequency of commas, suggesting a focus on lists or complex structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

²¾

-0.75

ãĥ¯ãĥ³

-0.74

ãģł

-0.70

itialized

-0.70

¶ħ

-0.69

)</

-0.68

cember

-0.68

etheless

-0.66

chnology

-0.64

(#

-0.64

POSITIVE LOGITS

 according

1.41

 says

1.35

 said

1.25

 experts

1.21

 writes

1.15

according

1.10

 explains

1.10

 argues

1.09

 analysts

1.06

 Says

1.05

Activations Density 0.503%

mentions of companies and organizations with a strong positive or negative association

sentences that include a high frequency of commas, suggesting a focus on lists or complex structures

No Comments

No Known Activations