INDEX

Explanations

responses in first-person narrative style

oai_token-act-pair · gpt-3.5-turbo

negations and expressions of limitations or inability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

CLR

-0.73

 Rosenstein

-0.66

 Lavrov

-0.63

Hof

-0.61

 Aerospace

-0.60

aic

-0.60

 Highlands

-0.60

idates

-0.59

ATF

-0.59

 Lange

-0.58

POSITIVE LOGITS

 myself

0.90

ngth

0.69

 praying

0.67

 liking

0.67

 personally

0.66

 reckoning

0.66

venge

0.65

oan

0.64

 writing

0.64

 regret

0.62

Activations Density 0.855%

responses in first-person narrative style

negations and expressions of limitations or inability

No Comments

No Known Activations