INDEX

Explanations

questions starting with "How do you" and "How do we" followed by verbs

oai_token-act-pair · gpt-3.5-turbo

query-based phrasing or questions regarding knowledge and understanding

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

pict

-0.66

letter

-0.62

iken

-0.61

tn

-0.61

 Boat

-0.61

letters

-0.61

prints

-0.60

Runner

-0.60

court

-0.59

requisite

-0.59

POSITIVE LOGITS

 reconcile

1.03

 reconcil

1.01

 cope

0.87

Ô

0.82

 differentiate

0.81

 distinguish

0.80

 handle

0.75

 justify

0.74

 navigate

0.73

 decipher

0.73

Activations Density 0.131%

questions starting with "How do you" and "How do we" followed by verbs

query-based phrasing or questions regarding knowledge and understanding

No Comments

No Known Activations