INDEX

Explanations

conjunctions and short phrases indicating a sequence of actions or events

oai_token-act-pair · gpt-3.5-turbo

phrases related to emotional reactions and interpersonal dynamics

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

 Compatibility

-0.60

 Create

-0.57

%%

-0.56

Join

-0.56

CIS

-0.55

 Locations

-0.54

tains

-0.54

 nutshell

-0.53

%%

-0.53

 Spread

-0.53

POSITIVE LOGITS

 confessed

1.14

 remarked

1.13

 apologised

1.13

 joked

1.11

 muttered

1.08

 replied

1.06

 recalled

1.05

 apologized

1.04

 insisted

1.04

 thanked

1.03

Activations Density 0.729%

conjunctions and short phrases indicating a sequence of actions or events

phrases related to emotional reactions and interpersonal dynamics

No Comments

No Known Activations