INDEX

Explanations

words containing the substring "ns" with a high level of activation

oai_token-act-pair · gpt-3.5-turbo

the presence of the word "ns" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

MAD

-0.68

gri

-0.64

 plac

-0.63

 blast

-0.59

 Madness

-0.59

 adjustment

-0.59

 freeze

-0.59

 absentee

-0.58

 fine

-0.57

 starship

-0.57

POSITIVE LOGITS

ns

4.57

ns

1.73

ls

1.52

nces

1.52

NS

1.50

nc

1.49

1.48

nt

1.43

nn

1.37

nes

1.35

Activations Density 0.010%

words containing the substring "ns" with a high level of activation

the presence of the word "ns" in various contexts

No Comments

No Known Activations