INDEX

Explanations

phrases related to legal or official actions

oai_token-act-pair · gpt-3.5-turbo

instances or phrases indicating intensity or magnitudes 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 @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

hement

-0.93

ataka

-0.74

ichick

-0.67

 diving

-0.66

romeda

-0.65

 withdrawal

-0.63

 undermin

-0.63

 unsuspecting

-0.63

 overt

-0.62

 charm

-0.61

POSITIVE LOGITS

ÂłÂłÂłÂł

0.90

SPONSORED

0.90

âĢ

0.88

Posted

0.88

âĹı

0.87

ÂłÂłÂłÂłÂłÂłÂłÂł

0.86

https

0.86

Trivia

0.85

Anonymous

0.84

################################

0.81

Activations Density 0.852%

phrases related to legal or official actions

instances or phrases indicating intensity or magnitudes in various contexts

No Comments

No Known Activations