INDEX

Explanations

action verbs related to people's roles or activities in various positions or organizations

oai_token-act-pair · gpt-3.5-turbo

actions and roles that individuals hold or perform

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

 Meaning

-0.60

itiveness

-0.57

!:

-0.56

 disadvant

-0.54

 manageable

-0.54

 Definitions

-0.53

 Ratings

-0.52

ifact

-0.52

 clich

-0.50

 misses

-0.49

POSITIVE LOGITS

 principally

0.89

sole

0.85

 alongside

0.82

 unsuccessfully

0.81

 primarily

0.81

 mainly

0.76

 chiefly

0.74

 exclusively

0.74

 solely

0.74

 jointly

0.72

Activations Density 0.309%

action verbs related to people's roles or activities in various positions or organizations

actions and roles that individuals hold or perform

No Comments

No Known Activations