Beta
Explore
Marketplace
Neural Labs
Playground
Wallet
Docs
qwen3-8b-speculators – AI Model by liujonathan24 | AlphaNeural AI
You can deploy this model and start earning money today!
liujonathan24
/
qwen3-8b-speculators
like
0
us
Views
No views yet
Model card
Files and Versions
Community
API
Deploy
Qwen3-8B Speculators (DFlash & FLM)
Speculator models for the
Qwen/Qwen3-8B
target model.
Architectures
Model File
mlp_type
Draft Layers
Style
dflash_5layer_mix.pt
dflash
5
DFlash-style
flm_5layer_uni.pt
flm
5
FLM-style
Training Data
These models were trained on the
regenerated dataset
(DFlash-style) derived from Qwen3-8B's own responses to the following sources:
Nemotron Post-Training V2
CodeAlpaca
ShareGPT
UltraChat-200K
Specifically:
dflash_5layer_mix
: Uses a mixture of datasets and a custom Tau schedule (30% pure noise, 20% grid-aligned, 50% uniform).
flm_5layer_uni
: Uses a uniform Tau schedule and standard dataset mixture.
Details
Target Model
: Qwen/Qwen3-8B
Sequence Length
: 2048
Batch Size
: 5
Draft Layers
: 5 (despite some folder naming suggesting 4, internal config confirms 5)
Trained at Princeton AILab.
