INDEX

Explanations

terms related to belief or trust

oai_token-act-pair · gpt-3.5-turbo

variations of the word "believe."

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

 design

-0.73

 background

-0.71

 cropped

-0.67

 snaps

-0.65

 supers

-0.64

 rehe

-0.61

 Zoro

-0.61

 designs

-0.61

 references

-0.60

 subt

-0.60

POSITIVE LOGITS

ieve

4.37

ieved

3.49

ieves

3.35

ieving

3.26

ievers

2.72

iever

2.60

ief

1.79

iev

1.64

oup

1.19

chieve

1.17

Activations Density 0.020%

terms related to belief or trust

variations of the word "believe."

No Comments

No Known Activations