Add base models for Employee agents

paper/conference_101719.tex

@@ -119,6 +119,7 @@ The system uses a two-tiered hierarchy:
             \item Task Execution: Receive task, execute, return result.
             \item Resource Consumption: Associated costs (API, memory) tracked by system.
         \end{itemize}
+        Specialized employee agents are constructed using base models such as Mistral~7B~\cite{jiang2023mistral}, Llama~3~\cite{llama3herd}, Gemini~1.5~\cite{gemini1.5_report}, Qwen2.5~\cite{qwen2.5_report}, Qwen3~\cite{qwen3_blog}, and DeepSeek-R1~\cite{deepseekr1_report}, with the CEO agent configuring them via tailored system prompts.
 \end{itemize}
 This hierarchy facilitates task decomposition and result aggregation; the dynamic pool provides flexibility.
 

paper/references.bib

@@ -394,3 +394,63 @@
       url={https://arxiv.org/abs/2407.03978}, 
 }
 
+@article{jiang2023mistral,
+  title={{Mistral 7B}},
+  author={Jiang, Albert Q and Xu, Alexandre and Lachaux, Arthur Mensch Guillaume Lample Nicol{\`a}s and Rozenberg, Fran{\c{c}}ois and Lacroix, Timoth{\'e}e and Lavril, Thibaut and Gaddipati, Teven Le Scao Eleonora and Ortiz, Lucile Saulnier Lixin and Tang, Dieuwke Hiemstra L{\'e}lio Renard and others},
+  year={2023},
+  eprint={2310.06825},
+  archivePrefix={arXiv},
+  primaryClass={cs.CL},
+  url={https://arxiv.org/abs/2310.06825},
+}
+
+@article{llama3herd,
+  title={{The Llama 3 Herd of Models}},
+  author={{Meta Llama Team}},
+  year={2024},
+  eprint={2407.21783},
+  archivePrefix={arXiv},
+  primaryClass={cs.CL},
+  url={https://arxiv.org/abs/2407.21783},
+  note={arXiv:2407.21783}
+}
+
+@article{gemini1.5_report,
+  title={{Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context}},
+  author={{Gemini Team}},
+  year={2024},
+  eprint={2403.05530},
+  archivePrefix={arXiv},
+  primaryClass={cs.CL},
+  url={https://arxiv.org/abs/2403.05530},
+  note={arXiv:2403.05530}
+}
+
+@article{qwen2.5_report,
+  title={{Qwen2.5 Technical Report}},
+  author={{Qwen Team} and Yang, An and others},
+  year={2024},
+  eprint={2412.15115},
+  archivePrefix={arXiv},
+  primaryClass={cs.CL},
+  url={https://arxiv.org/abs/2412.15115},
+  note={arXiv:2412.15115}
+}
+
+@misc{qwen3_blog,
+    title={{Qwen3: Think Deeper, Act Faster}},
+    author={{Qwen Team}},
+    howpublished={\url{https://qwenlm.github.io/blog/qwen3/}},
+    year={2025}
+}
+
+@article{deepseekr1_report,
+  title={{DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning}},
+  author={{DeepSeek-AI and others}},
+  year={2025},
+  eprint={2501.12948},
+  archivePrefix={arXiv},
+  primaryClass={cs.CL},
+  url={https://arxiv.org/abs/2501.12948},
+  note={arXiv:2501.12948}
+}