Downloads · 30 days
12
36% of all-time downloads
AkanaYB/saas-erp-support-qwen3-4b-adapter-f
saas-erp-support-qwen3-4b-adapter-f is a text generation model from AkanaYB. Use it when you need the model to write or continue text. It is set up for peft.
Experimental Russian SaaS / fictional ERP / telecom support LoRA adapter. Published project artifact; project-specific licensing has not been selected. The upstream Base license does not automatically license the auth…
Downloads · 30 days
12
36% of all-time downloads
All-time downloads
33
Public
Repo size
144 MB
Likes
0
Public
Click a slice to open those files.
.safetensors132 MB · 89%
From the Hugging Face model README
Experimental Russian SaaS / fictional ERP / telecom support LoRA adapter. Published project artifact; project-specific licensing has not been selected. The upstream Base license does not automatically license the authored adapter or data. Code and launch tutorial: https://github.com/Akana92/SaaS-ERP-CRM-Support-Client
Base: Qwen/Qwen3-4B-Instruct-2507, revision
cdbee75f17c01a7cc42f958dc650907174af0554 (download separately; no base weights here).
Clean-base QLoRA NF4, rank 16, alpha 32, dropout .05, all linear layers,
two epochs / 558 optimizer steps. Training set: 430 synthetic train dialogues,
1113 target turns, 13 local shards. Pending human review.
Use the project's pinned runtime and PEFT loader with the adapter files here;
the copied adapter configuration points to the pinned Hub base instead of a local path.
Loading this adapter alone does not reproduce LangGraph, ERP context, or server decisions.
AI-assessed development comparison: Base 10/100 vs F 49/100 fully successful
dialogues, 225 turns, NF4. Status: QUALIFIED. Success requires the five labels
and semantic response criteria across all turns. This is not human-verified or
final-test accuracy: human review is absent, final sealed test was not run,
full_success_rate=null. Live BF16 screenshots are not a new evaluation.
Do not claim 90% quality or production readiness. Real ERP integration, production
authorization, and measured operator time savings are not provided.
BASE_MODEL_LICENSE is the unchanged Apache-2.0 license from the pinned Qwen
base snapshot. It applies to upstream Qwen assets (including copied tokenizer
assets), and is not a license grant for this project's adapter. Upstream source:
https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507/tree/cdbee75f17c01a7cc42f958dc650907174af0554
manifest.json records source and export hashes. Only the COPY of
adapter_config.json changes base_model_name_or_path and revision; originals stay frozen.
No optimizer states, checkpoints, evaluation questions, gold labels, raw reviews,
isolation receipt, or secrets are included. This is not a complete experimental
reproduction bundle. See the code repository's docs/ML_ASSETS.md.