INDEX

Explanations

instances where something has been effective or successful

oai_token-act-pair · gpt-3.5-turbo

terms related to the effectiveness of strategies or actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ortment

-0.73

ewitness

-0.72

ritical

-0.71

antha

-0.69

vv

-0.67

involved

-0.66

pora

-0.66

-0.65

ailable

-0.63

estern

-0.61

POSITIVE LOGITS

 wonders

1.27

 miracles

0.99

 miser

0.95

 brilliantly

0.95

 well

0.90

 spectacular

0.85

 satisf

0.85

heet

0.85

 beautifully

0.84

 against

0.83

Activations Density 0.120%

instances where something has been effective or successful

terms related to the effectiveness of strategies or actions

No Comments

No Known Activations