INDEX

Explanations

terms related to themes or specific elements within a larger context

oai_token-act-pair · gpt-3.5-turbo

word patterns related to themes, courses, and scenes in structured formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 senate

-0.99

 navy

-0.86

 department

-0.85

 hall

-0.85

 dwar

-0.85

 brigade

-0.84

 academy

-0.84

 station

-0.81

 glor

-0.81

 stake

-0.81

POSITIVE LOGITS

 Course

1.87

 Error

1.62

 Theme

1.61

 Address

1.58

 Template

1.58

 Languages

1.58

 Format

1.57

 Position

1.56

 Account

1.55

 Pattern

1.55

Activations Density 0.135%

terms related to themes or specific elements within a larger context

word patterns related to themes, courses, and scenes in structured formats

No Comments

No Known Activations