INDEX

Explanations

specific instructions or prohibitions

oai_token-act-pair · gpt-3.5-turbo

phrases containing the word "that" in various contexts, indicating a focus on specifying conditions or details

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

————

-0.61

 whisk

-0.60

ointed

-0.57

apsed

-0.57

ahime

-0.55

eely

-0.54

 etched

-0.54

 realizing

-0.54

emon

-0.54

Hope

-0.53

POSITIVE LOGITS

 violates

1.23

 exceeds

1.21

 involves

1.16

 contradicts

1.14

 occurs

1.08

 disagrees

1.08

 qualifies

1.06

isn

1.05

 doesn

1.05

 satisfies

1.01

Activations Density 0.156%

specific instructions or prohibitions

phrases containing the word "that" in various contexts, indicating a focus on specifying conditions or details

No Comments

No Known Activations