INDEX

Explanations

mentions of a specific name or surname "Lombard"

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of the term "Lombard" or variations of it

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

 icing

-0.74

 commentary

-0.67

 saturation

-0.61

paw

-0.61

 peanuts

-0.59

Paw

-0.58

Pwr

-0.58

 almonds

-0.58

gou

-0.58

 displacement

-0.56

POSITIVE LOGITS

rero

1.27

ardy

1.15

ucha

1.15

oys

1.08

osi

1.07

ritis

1.07

uds

1.04

stones

1.04

reon

1.02

rows

1.02

Activations Density 0.047%

mentions of a specific name or surname "Lombard"

repeated mentions of the term "Lombard" or variations of it

No Comments

No Known Activations