INDEX

Explanations

programming code snippets

oai_token-act-pair · gpt-3.5-turbo

sequences of colons and patterns indicative of programming syntax

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

 business

-0.65

 Semin

-0.65

egg

-0.65

 Ashe

-0.65

 Loud

-0.61

 Mang

-0.61

 drills

-0.59

 Sanford

-0.59

 Simmons

-0.58

 drill

-0.57

POSITIVE LOGITS

::

3.88

::

2.36

::::

1.81

->

1.47

=>

1.39

::::::::

1.27

++

1.10

*)

1.10

():

1.09

Â®

1.05

Activations Density 0.015%

programming code snippets

sequences of colons and patterns indicative of programming syntax

No Comments

No Known Activations