FreePublished on GitHub

chinese-llama-alpaca

Route Chinese-LLaMA-Alpaca model reconstruction, inference/deployment, training/fine-tuning, and evaluation workflows.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill

Works withOpenAI

Run score Not tested

See how to install

Files included (54)

SKILL.md4.2 KB
references/compliance-and-limitations.md1.5 KB
references/environment-and-installation.md2.1 KB
references/model-family-overview.md2.0 KB
references/repo-provenance.md1.7 KB
references/repo-routing-metadata.json564 B
references/troubleshooting.md2.6 KB
scripts/check_environment.py3.7 KB
scripts/verify_sha256.py2.1 KB
sub-skills/evaluation-benchmarks/references/ceval-workflow.md2.8 KB
sub-skills/evaluation-benchmarks/references/example-benchmarks.md3.0 KB
sub-skills/evaluation-benchmarks/references/troubleshooting.md2.8 KB
sub-skills/evaluation-benchmarks/scripts/ceval/eval.py5.9 KB
sub-skills/evaluation-benchmarks/scripts/ceval/evaluator.py1.5 KB
sub-skills/evaluation-benchmarks/scripts/ceval/llama_evaluator.py8.1 KB
sub-skills/evaluation-benchmarks/scripts/ceval/subject_mapping.json5.2 KB
sub-skills/evaluation-benchmarks/scripts/validate_ceval_layout.py3.9 KB
sub-skills/evaluation-benchmarks/SKILL.md2.6 KB
sub-skills/inference-deployment/references/gradio-and-api-server.md3.5 KB
sub-skills/inference-deployment/references/hf-inference.md3.7 KB
sub-skills/inference-deployment/references/langchain-integrations.md2.7 KB
sub-skills/inference-deployment/references/troubleshooting.md4.0 KB
sub-skills/inference-deployment/scripts/doc.txt6.0 KB
sub-skills/inference-deployment/scripts/gradio_demo.py11.1 KB
sub-skills/inference-deployment/scripts/inference_hf.py7.1 KB
sub-skills/inference-deployment/scripts/langchain_qa.py4.2 KB
sub-skills/inference-deployment/scripts/langchain_sum.py2.7 KB
sub-skills/inference-deployment/scripts/openai_api_protocol.py2.2 KB
sub-skills/inference-deployment/scripts/openai_api_server.py7.8 KB
sub-skills/inference-deployment/scripts/patches.py8.4 KB
sub-skills/inference-deployment/SKILL.md3.1 KB
sub-skills/model-reconstruction/references/checksums.md3.6 KB
sub-skills/model-reconstruction/references/model-assets.md3.8 KB
sub-skills/model-reconstruction/references/reconstruction-workflows.md3.7 KB
sub-skills/model-reconstruction/references/troubleshooting.md3.9 KB
sub-skills/model-reconstruction/scripts/chinese_sp.model507.8 KB
sub-skills/model-reconstruction/scripts/merge_llama_with_chinese_lora_low_mem.py14.5 KB
sub-skills/model-reconstruction/scripts/merge_llama_with_chinese_lora.py14.4 KB
sub-skills/model-reconstruction/scripts/merge_tokenizers.py2.8 KB
sub-skills/model-reconstruction/SKILL.md3.1 KB
sub-skills/training-finetuning/references/api-reference.md7.7 KB
sub-skills/training-finetuning/references/data-formats.md5.8 KB
sub-skills/training-finetuning/references/training-workflows.md8.6 KB
sub-skills/training-finetuning/references/troubleshooting.md6.8 KB
sub-skills/training-finetuning/scripts/build_dataset.py3.9 KB
sub-skills/training-finetuning/scripts/run_clm_pt_with_peft.py27.6 KB
sub-skills/training-finetuning/scripts/run_clm_sft_with_peft.py17.5 KB
sub-skills/training-finetuning/scripts/validate_training_data.py6.1 KB
sub-skills/training-finetuning/SKILL.md3.6 KB
sub-skills/training-finetuning/templates/ds_zero2_no_offload.json686 B
sub-skills/training-finetuning/templates/instruction_sample.json372 B
sub-skills/training-finetuning/templates/pretrain_sample.txt221 B
sub-skills/training-finetuning/templates/run_pt.sh3.3 KB
sub-skills/training-finetuning/templates/run_sft.sh3.5 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Chinese-LLaMA-Alpaca Repo Skill

Use this repo skill when a task involves the original Chinese-LLaMA-Alpaca project: Chinese LLaMA/Alpaca LoRA adapters, tokenizer expansion, LoRA reconstruction, local Transformers inference, Gradio or OpenAI-compatible serving, LangChain demos, PEFT pretraining/SFT, C-Eval, or the repo's example benchmark tables. The command examples below assume the current working directory is this generated skill root.

This skill is self-contained for workflow guidance and bundled helper scripts, but it does not provide original LLaMA weights, Chinese LoRA downloads, C-Eval data, API keys, or permission to use restricted assets commercially. Always confirm asset/license, hardware, network, service, and budget constraints before running heavyweight commands.

Start Here

Read references/repo-provenance.md when checking freshness against a repository checkout.
Read references/model-family-overview.md to choose Chinese LLaMA versus Chinese Alpaca, size, Plus/Pro, and tokenizer family.
Read references/environment-and-installation.md before creating an environment or installing optional dependencies.
Run scripts/check_environment.py for a safe dependency/backend probe.
Use scripts/verify_sha256.py with the checksum reference when checking downloaded model/tokenizer assets.
Use references/troubleshooting.md for cross-cutting install/import/asset/backend failures.
Use references/compliance-and-limitations.md before publishing outputs or advising on commercial/safety-sensitive use.

Route Map

User request	Use
"Merge Chinese Alpaca LoRA with LLaMA", "convert to HF/PTH", "tokenizer mismatch", "verify adapter checksum"	`sub-skills/model-reconstruction/`
"Run inference", "batch predictions", "interactive single-turn", "Gradio demo", "OpenAI API server", "LangChain QA"	`sub-skills/inference-deployment/`
"Prepare SFT data", "validate instruction JSON", "pretrain Chinese LLaMA", "fine-tune Alpaca LoRA", "DeepSpeed/PEFT args"	`sub-skills/training-finetuning/`
"Run C-Eval", "validate C-Eval data", "interpret examples scores", "compare q4/q8/Plus/Pro"	`sub-skills/evaluation-benchmarks/`

Minimal Public Environment Check

The original requirements.txt pins:

torch==1.13.1
transformers==4.30.0
sentencepiece==0.1.97
PEFT from the repository's pinned Hugging Face commit

Additional workflows need optional packages such as datasets, pandas, scikit-learn, fastapi, uvicorn, shortuuid, gradio, langchain, FAISS, or DeepSpeed. Install only the optional group required by the selected workflow.

Safe check from the generated skill root:

python scripts/check_environment.py --include-optional

Use python scripts/verify_sha256.py /path/to/file --expected name=hex from the same skill root when validating downloaded model, tokenizer, or data files. These helpers only import packages, check CUDA visibility, or hash files; they do not download models, launch servers, run training, or read credentials.

Common Boundaries

If a user only has LoRA files and wants generation, reconstruct or load the LoRA with a compatible base model first.
If a user wants chat/instruction following, prefer Chinese Alpaca and use the Alpaca prompt template. Chinese LLaMA is base/continuation-oriented.
If a user wants training, validate data before allocating GPUs. Real training is long-running.
If a user wants benchmark claims, record model path, dataset, decoding flags, hardware, and skipped/failed cases; do not infer global quality from example tables.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

More skills from this repository

3d-resnets-pytorch

Free

Routes 3D ResNets PyTorch video action-recognition workflows across training, inference, and data preparation.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

3ddfa

Free

Guide 3DDFA Python inference, geometry rendering, training/evaluation, and optional C++ ONNX workflows for 3D dense face alignment.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

3ddfa-v2

Free

Routes 3DDFA_V2 face-alignment setup, still-image demos, video tracking, and ONNX benchmarking workflows.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

ab3dmot

Free

Operate AB3DMOT 3D multi-object tracking workflows for KITTI and nuScenes data, tracking, evaluation, and visualization.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

accelerate

Free

Use Hugging Face Accelerate for PyTorch training-loop migration, distributed launch/configuration, DeepSpeed/FSDP/TPU backend setup, big-model inference/offload, checkpointing, tracking, and troubleshooting.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

acme

Free

Route Acme reinforcement-learning framework tasks across core loops, replay/data, JAX agents, and TensorFlow/Sonnet agents.

The English summary is being prepared. Showing the original description.

VectorSpaceLab/AREX-Skill341Updated Sep 3, 2026

See all skills by VectorSpaceLab

Report a problem with this skill

Install

Install with a command

npx skills add https://github.com/VectorSpaceLab/AREX-Skill --skill chinese-llama-alpaca

Paste it into your terminal and run it (requires Node.js). You can choose which agents to install it for, such as Claude Code or Codex.

Ask your agent

Please install the following Agent Skill.
https://github.com/VectorSpaceLab/AREX-Skill/tree/main/skills/repositories/repo-skills/chinese-llama-alpaca

First review SKILL.md and the bundled files and confirm there is nothing suspicious (such as sending data externally or reading credentials). Then place it in ~/.claude/skills/chinese-llama-alpaca/ for Claude Code, or ~/.agents/skills/chinese-llama-alpaca/ for Codex.

Send this text to Claude Code or Codex as is. It checks the contents before installing.

View source on GitHub

Repository: VectorSpaceLab/AREX-Skill
Stars: 341
Forks: 26
Category: DevOps
License: Apache-2.0
Original language: English
Last updated: Sep 3, 2026

A skill is a set of instructions and scripts for an agent. cccskills does not guarantee that skills are safe. Review the contents before you install.