初始化项目，由ModelHub XC社区提供模型

Model: yujianboisme/qwen2.5-1.5b-darkhorse-code-fine-tuning Source: Original Platform
2026-09-29 03:31:14 +08:00
commit 5b07c289be
70 changed files with 1284 additions and 0 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,50 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*.tfevents* filter=lfs diff=lfs merge=lfs -text
+*.db* filter=lfs diff=lfs merge=lfs -text
+*.ark* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*data* filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.meta filter=lfs diff=lfs merge=lfs -text
+**/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.gguf* filter=lfs diff=lfs merge=lfs -text
+*.ggml filter=lfs diff=lfs merge=lfs -text
+*.llamafile* filter=lfs diff=lfs merge=lfs -text
+*.pt2 filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+"/merges.txt" filter=lfs diff=lfs merge=lfs -text
+"/tokenizer.json" filter=lfs diff=lfs merge=lfs -text
+"/vocab.json" filter=lfs diff=lfs merge=lfs -text
--- a/README.md
+++ b/README.md
@@ -0,0 +1,319 @@
+---
+license: apache-2.0
+license_link: https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct/blob/main/LICENSE
+language:
+- zh
+pipeline_tag: text-generation
+library_name: transformers
+base_model: Qwen/Qwen2.5-1.5B-Instruct
+tags:
+- text-generation
+- qwen2.5
+- lora
+- json
+- instruction-following
+- chinese
+- command-translation
+---
+
+# darkhorse-code-instruct-1.5b
+
+> **把一个微调过的 1.5B 小模型，做成编辑器里的「自然语言 → 私域命令」翻译器。**
+> 输入一句中文口语，输出**一行严格 JSON**，不是聊天回复。
+
+```
+输入：关掉倒数第5个文件
+输出：{"intent": "command", "cmd": "close -5"}
+```
+
+```json
+{"intent": "command", "cmd": "del src/P05-es/pdf2es.py"}                 // 删除类
+{"intent": "reject",  "reply": "您没有权限！"}                             // 越权/不合法
+{"intent": "chat",    "reply": "你还是好好工作吧，房贷还清了吗？车贷还清了吗？"}  // 非命令输入
+```
+
+本模型是 **Qwen2.5-1.5B-Instruct + LoRA（r=16）** 指令微调后**合并**（merged）得到的自包含权重，
+不需要 `peft`、不需要 adapter，`transformers` / vLLM 直接加载即可。
+
+---
+
+## 1. 它解决什么问题
+
+`darkhorse-code` 是一个编辑器项目，它有一套**私域命令语法**（`open project` / `close -3` /
+`config add -r k=v` …），命令语法固定、可校验，但用户不想背。于是把「用户口语 → 一条标准命令」
+这个窄任务交给一个 1.5B 小模型：**小、快、可完全本地跑，且输出必须能被程序解析**。
+
+它不是通用聊天模型，**不会**回答知识问题、不会写代码、不会陪你聊天 ——
+遇到非命令输入，它的正确行为是回一句 20 字以内的短回复（或一句固定的「劝退」话术）。
+
+## 2. 输出契约（三种信封）
+
+模型**只输出一行 JSON**，`intent` 决定信封形状：
+
+| intent | 形状 | 含义 |
+|---|---|---|
+| `command` | `{"intent":"command","cmd":"<标准命令>"}` | 翻译成功，`cmd` 可直接执行 |
+| `reject` | `{"intent":"reject","reply":"<拒绝原因>"}` | 命中硬性规则，如路径越权 |
+| `chat` | `{"intent":"chat","reply":"<≤20字短回复>"}` | 不是私域命令 |
+
+两条固定话术（评测按**逐字**判定）：
+
+- 绝对路径出现在 `open file` / `new` / `del` / `rename` → `您没有权限！`
+- 要求把运行目标存成全局配置 → `运行目标不能保存为全局`
+- 闲聊超过 20 字说不完 → `你还是好好工作吧，房贷还清了吗？车贷还清了吗？`
+
+## 3. 命令语法（模型学到的目标空间）
+
+```
+open project <绝对路径> | open file <相对路径>
+close project|all|other|left|right|<序号>          # 序号 0 起，负数从右往左，-1 是最后一个
+config add|remove|update|get [-g|-p|-r] <key>=<value>   # 未指定层级默认 -p
+new file <相对路径> | new folder <相对路径> | new py|rs|md|c <名称>
+del <相对路径> | run <名称>=<命令> | run del <名称>
+rename <旧相对路径> <新相对路径>
+project lang <语言> <绝对路径> | project delete <绝对路径> | project migrate
+project edit "<路径>" "<名称>" <语言>
+```
+
+语言取值：`unknown/mix/java/c/python/rust/web/golang/document/kotlin`（中文表达做了映射：前端→web、Go→golang、文档→document…）。
+
+## 4. 快速开始（transformers）
+
+⚠️ **两条铁律，违反任一条效果都会明显掉：**
+
+1. **必须使用本模型内置的 system prompt**（下面 `SYSTEM_PROMPT` 原文，逐字），
+   模型是在这份 prompt 下微调的；换 prompt = 训练/推理不一致。
+2. **必须走模型自带的 chat template**（本仓库的 `chat_template.jinja`），
+   `tokenizer.apply_chat_template` 会自动读到它。
+
+```python
+import json
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+MODEL = "darkhorse-code-instruct-1.5b"   # 换成你的 ModelScope 仓库 id
+tok = AutoTokenizer.from_pretrained(MODEL)
+model = AutoModelForCausalLM.from_pretrained(MODEL, torch_dtype="bfloat16", device_map="auto")
+
+SYSTEM_PROMPT = """你是 darkhorse-code 编辑器的命令翻译器。把用户的话翻译成一条标准命令，只输出一行 JSON。
+
+命令语法：
+open project <绝对路径> | open file <相对路径>
+close project|all|other|left|right|<序号>
+close 的序号从 0 开始，负数从右往左，-1 是最后一个
+config add|remove|update|get [-g|-p|-r] <key>=<value>，未指定层级默认 -p
+new file <相对路径> | new folder <相对路径> | new py|rs|md|c <名称>
+del <相对路径> | run <名称>=<命令> | run del <名称>
+rename <旧相对路径> <新相对路径>
+project lang <语言> <绝对路径> | project delete <绝对路径> | project migrate
+project edit "<路径>" "<名称>" <语言>
+
+规则：
+1. 语言取值 unknown/mix/java/c/python/rust/web/golang/document/kotlin；中文表达要映射：Python→python、Go→golang、前端→web、文档→document、混合→mix、未知→unknown。
+2. 绝对路径（以 / 或 C:/ D:/ 开头）只允许用在 open project、project lang/edit/delete 和 config 的值里；出现在 open file、new、del、rename 时一律拒绝，reply 固定为「您没有权限！」。
+3. 路径里的正斜杠输出一律转成反斜杠。
+4. 运行目标只能存项目配置，不能存全局；用户要求存全局时 reply 固定为「运行目标不能保存为全局」。
+5. config 的值含空格时用双引号包裹。上下文给出已有运行目标时，新增运行目标索引 = 最大索引 + 1。
+6. 不是私域命令的输入（闲聊、问知识、要你写文档）用 chat：reply 不超过 20 字；一句话说不完 20 字时 reply 固定为「你还是好好工作吧，房贷还清了吗？车贷还清了吗？」。
+7. 同义命令输出规范形式：删除类写 del，关闭其他文件写 close other，重命名写 rename，新建目录写 new folder，run 的删除写 run del。
+
+输出格式（只输出一行 JSON，不要解释、不要代码块）：
+{"intent":"command","cmd":"<标准命令>"}
+{"intent":"reject","reply":"<拒绝原因>"}
+{"intent":"chat","reply":"<短回复>"}"""
+
+
+def translate(text: str, context: str | None = None) -> dict:
+    # 上下文按训练时的约定另起一行：'[上下文] <内容>'
+    user = f"[上下文] {context}\n{text}" if context else text
+    msgs = [{"role": "system", "content": SYSTEM_PROMPT},
+            {"role": "user", "content": user}]
+    prompt = tok.apply_chat_template(msgs, tokenize=False, add_generation_prompt=True)
+    ids = tok(prompt, return_tensors="pt", add_special_tokens=False).to(model.device)
+    out = model.generate(**ids, max_new_tokens=128, do_sample=False,   # 贪心，别采样
+                         pad_token_id=tok.pad_token_id or tok.eos_token_id)
+    raw = tok.decode(out[0][ids["input_ids"].shape[1]:], skip_special_tokens=True).strip()
+    return json.loads(raw)          # 建议外面再包一层 try：小模型偶尔会多写一句解释
+
+
+print(translate("把倒数第三个文件关掉"))
+# {'intent': 'command', 'cmd': 'close -3'}
+print(translate("加一个运行目标 测试=pytest -q", context="当前项目: D:/Projects/demo"))
+# {'intent': 'command', 'cmd': 'run 测试=pytest -q'}
+```
+
+## 5. vLLM 部署（推荐，OpenAI 兼容）
+
+8 GB 显存（如 RTX 5060 Laptop）即可跑 bf16：
+
+```bash
+vllm serve ./qwen2.5-1.5b-cmd-merged \
+  --served-model-name darkhorse-code-instruct \
+  --chat-template ./qwen2.5-1.5b-cmd-merged/chat_template.jinja \
+  --dtype bfloat16 --max-model-len 32768 \
+  --gpu-memory-utilization 0.85 --max-num-seqs 32 \
+  --enable-prefix-caching --port 8001
+```
+
+```bash
+curl http://127.0.0.1:8001/v1/chat/completions -H "Content-Type: application/json" -d '{
+  "model": "darkhorse-code-instruct",
+  "messages": [
+    {"role": "system", "content": "<上面那份 SYSTEM_PROMPT 原文>"},
+    {"role": "user", "content": "关掉倒数第5个文件"}
+  ],
+  "temperature": 0, "max_tokens": 128
+}'
+```
+
+要点：
+
+- `temperature=0`（训练与评测都是贪心解码）；**不要**用 `generation_config.json` 里的默认采样参数。
+- 请求里的 `system` 必须是上面那份原文；本项目服务端做薄封装时会把 system 强制替换成它。
+- `choices[0].message.content` 是**契约 JSON 字符串**，调用方需要 `json.loads`。
+- `--enable-prefix-caching` 对这类场景收益明显：所有请求共享同一份 553 token 的 system prompt 前缀。
+- 需要「输出结构绝对合法」时，可以上受约束解码（JSON Schema 限死三种信封），
+  代价是采样空间被约束，个别内容可能与贪心输出不同。
+
+### 实测性能（RTX 5060 Laptop 8G / sm_120，bf16）
+
+同一批 15 条真实指令压测，`--max-model-len 32768 --max-num-seqs 32 --enable-prefix-caching`：
+
+| 并发 | 请求吞吐 | 生成吞吐 | 延迟 p50 | 延迟 p95 |
+|---|---|---|---|---|
+| 1 | 3.6 req/s | 72 tok/s | 251 ms | 389 ms |
+| 8 | 23.0 req/s | 469 tok/s | 278 ms | 425 ms |
+| 32 | **49.4 req/s** | **1002 tok/s** | 337 ms | 480 ms |
+
+流式 TTFT（并发 1）**87 ms**；显存占用：权重 3.1 G + KV cache 3.46 GiB（129,680 token ⇒ 32K 上下文可并发 3.96 路）。
+参考：同一模型用 CPU + transformers 逐条生成是 **4.1–6.4 s/条**（约 0.2 req/s）。
+
+### 在 WSL2 上部署的三个额外环境变量
+
+vLLM 官方只发 Linux 轮子，Windows 用户通常跑在 WSL2 里；此时还需要：
+
+```bash
+# 1) V2 Model Runner 需要 pinned memory/UVA，而 vLLM 在 WSL2 上默认关掉 pin memory
+export VLLM_WSL2_ENABLE_PIN_MEMORY=1
+# 2) WSL 里若没有 gcc，Triton 运行时编译 kernel launcher 会报 "Failed to find C compiler"
+#    可装 gcc（sudo apt-get install -y gcc g++），或用一个 zig cc 包装脚本当 CC
+export CC=/path/to/your/c-compiler
+# 3) 默认的 FlashInfer 采样器在缺 cubin 时会 JIT（需要 nvcc）；无 CUDA toolkit 时关掉它
+export VLLM_USE_FLASHINFER_SAMPLER=0
+```
+
+另外 WSL2 无 CUDA toolkit 时建议 `--compilation-config '{"mode":0}'`（跳过 torch.compile，启动更快、不需要 C++ 编译器）。
+
+## 6. 评测
+
+评测集 **131 条**，全部**未出现在训练集模板**中的说法（含 27 条刻意构造的硬样本）：
+中文序数词+负索引、口语化包装（「帮我/麻烦/能不能」）、上下文注入（`[上下文] 当前项目: …`）、
+同义命令归一（删除/删掉/移除 → `del`）、绝对路径拒绝、运行目标索引推断…
+
+| 指标 | 基座 Qwen2.5-1.5B-Instruct | **本模型（LoRA 微调后）** |
+|---|---|---|
+| 输出是合法 JSON | 131/131 = 100% | **131/131 = 100%** |
+| 与标准命令**完全一致**（exact） | 39/131 = 29.8% | **122/131 = 93.1%** |
+
+> 基座 JSON 合法率也是 100%，是因为语法和输出格式就写在 system prompt 里 ——
+> 但「知道语法」≠「会用语法」：55 个命令族里有 39 个族的准确率发生变化，
+> 基座在序号、配置、闲聊、拒绝这些地方大面积翻车。
+
+几族的对比（体现微调到底学到了什么）：
+
+| 命令族 | 基座 | 本模型 | 说明 |
+|---|---|---|---|
+| `close_index_neg`（倒数第 N 个） | 0/7 | **7/7** | 中文序数词 → 负索引 |
+| `del`（删除/删掉/移除归一） | 5/9 | **9/9** | 同义命令规范形式 |
+| `new_typed`（`new py/rs/md/c` 省后缀） | 0/4 | **4/4** | 类型子命令 |
+| `run_add_ctx`（上下文推索引） | 0/5 | **5/5** | `max+1` 索引推断 |
+| `reject_abs_dir` / `reject_abs_file` | 0/4 / 0/3 | **4/4 / 3/3** | 固定拒绝话术 |
+| `open_project_fwd`（正斜杠归一） | 0/3 | **3/3** | `/` → `\` |
+| `chat_long`（超 20 字固定话术） | 0/3 | **3/3** | 逐字背下话术 |
+| `config_*`（add/get/update/remove） | 0/1~0/3 | **多数转正** | 仍是最弱的一族，见局限 |
+
+**未满分的族（诚实列出，共 8 个）**：`config_get` 2/3、`config_remove` 1/3、
+`hard/config` 2/3、`hard/close_index` 1/2、`hard/open_project` 1/2、`hard/reject_abs` 1/2、
+`hard/reject_run_global` 0/1、`chat_short` 0/1。
+典型错误是**配置层级判断**：用户说「去掉配置项 X」时，模型输出 `config remove -p X`，
+而标准答案是 `config remove -r X`（这类「默认层级到底该取谁」的语义在数据里信号不够强）。
+
+## 7. 输入输出示例（**真实推理输出**，取自上述评测集）
+
+```
+输入：关掉倒数第5个文件
+输出：{"intent": "command", "cmd": "close -5"}
+
+输入：打开 C:/Users/yujia/PycharmProjects/ai-notes
+输出：{"intent": "command", "cmd": "open project C:\\Users\\yujia\\PycharmProjects\\ai-notes"}
+      （正斜杠已按规则归一为反斜杠）
+
+输入：删除 C:/Users/yujia/PycharmProjects/tmf/src/utils.py
+输出：{"intent": "reject", "reply": "您没有权限！"}
+      （绝对路径出现在 del 上 → 固定拒绝）
+
+输入：[上下文] 已有运行目标: target3
+      加个运行目标 测试=.venv\Scripts\python.exe -m uvicorn app:api --port 8000
+输出：{"intent": "command", "cmd": "run 测试=.venv\\Scripts\\python.exe -m uvicorn app:api --port 8000"}
+
+输入：把 darkhorse.code.editor.tab_size 设置成 4，存到运行时
+输出：{"intent": "command", "cmd": "config add -r darkhorse.code.editor.tab_size=4"}
+
+输入：帮我建一个 rust 的模块，叫 order_service
+输出：{"intent": "command", "cmd": "new rs order_service"}
+
+输入：D:\Projects\Python\Code-Rag 是Kotlin项目
+输出：{"intent": "command", "cmd": "project lang kotlin D:\\Projects\\Python\\Code-Rag"}
+
+输入：你觉得我该不该辞职去创业，帮我分析一下利弊
+输出：{"intent": "chat", "reply": "你还是好好工作吧，房贷还清了吗？车贷还清了吗？"}
+```
+
+## 8. 训练细节
+
+| 项 | 取值 |
+|---|---|
+| 基座 | `Qwen/Qwen2.5-1.5B-Instruct` |
+| 微调方式 | LoRA（`r=16`、`alpha=32`、`dropout=0.05`），训练后**合并**成自包含权重 |
+| 可训练参数 | 约 1.18%（其余冻结） |
+| 训练数据 | **1638 条**（模板 + 槽位合成，`--seed 42` 完全可复现，不用 LLM 生成） |
+| 超参 | `lr=2e-4`、`warmup_ratio=0.03`、2 epoch、`max_len=704`、`micro_bs=1` × `grad_accum=8` → 410 步 |
+| 耗时 / 显存 | 45.1 分钟 / 峰值 **7.68 GB**（单卡 8 GB 笔记本 GPU） |
+| loss | train 0.6818 → 0.0079（末轮均值 0.0319）；eval 0.039 → 0.0092 |
+| 精度 | bf16 |
+
+**数据为什么是合成的**：这个任务要学的是「**照抄**」而不是「记住」—— 命令里大量是路径、命令串、
+文件名。生成器把它们做成随机槽位，模型必须学会把输入里的路径原样搬进输出，而不是背样本；
+同时保证可复现、零 API 成本、不会把外部 LLM 的错误学进去。
+
+**三个刻意设计的难点**：① 约 15% 样本带上下文行，既要会用上下文补全、也要会忽略无关上下文；
+② 随机加「帮我/麻烦/能不能/顺手」前缀与「吧/，谢谢/呀」后缀，放大表达多样性；
+③ 27 条硬样本**只进评测集**，用来量真实泛化而不是量记忆。
+
+## 9. 已知局限（请务必先读）
+
+1. **只懂 darkhorse-code 的命令语**。换个 CLI/编辑器的命令体系，必须重新造数据微调。
+2. **不是聊天模型**：闲聊会被压成 ≤20 字短回复，甚至回那句固定的「劝退」话术 —— 这是设计目标，不是 bug。
+3. **`config` 族最弱**（见 §6）：配置层级（`-g/-p/-r`）的默认值判断会出错，接入方应对
+   `config` 类输出做二次校验（层级缺失时按自己的业务规则补默认值）。
+4. **输出偶尔会被解析层救回来**：小模型可能把 JSON 包在 ```json 代码块里、或路径里写单反斜杠
+   （非法 JSON 转义）。建议接入方保留「抠 JSON + 修非法转义」的解析兜底，而不是直接 `json.loads`。
+5. **上下文格式是约定**：必须写成 `[上下文] <内容>` 并**另起一行**再接用户原话，
+   这是训练时的格式，别自由发挥。
+6. **1.5B 的常识/推理上限**：超出命令翻译的语义理解（比如反讽、多轮澄清）不要指望它。
+
+## 10. 文件说明
+
+| 文件 | 说明 |
+|---|---|
+| `model.safetensors` | bf16 合并权重（约 2.9 GB），自包含、无需 adapter |
+| `chat_template.jinja` | **模型配套的 chat template**，请务必用它（vLLM `--chat-template` 指向它） |
+| `config.json` / `generation_config.json` | Qwen2 结构（28 层、2 KV head、`max_position_embeddings=32768`） |
+| `tokenizer.json` / `tokenizer_config.json` / `vocab.json` / `merges.txt` / `added_tokens.json` | tokenizer 全套 |
+| `configuration.json` | ModelScope 标记文件 |
+
+## 11. 许可与致谢
+
+- 本模型以 **Apache-2.0** 许可发布，与基座 `Qwen/Qwen2.5-1.5B-Instruct` 一致；
+  使用前请同时遵守基座模型的许可条款。
+- 感谢 Qwen 团队开源的 Qwen2.5 系列。
+- 训练所用数据为由模板与槽位**程序生成**的合成数据，不含任何真实用户数据或第三方隐私内容。
--- a/added_tokens.json
+++ b/added_tokens.json
@@ -0,0 +1,24 @@
+{
+  "</tool_call>": 151658,
+  "<tool_call>": 151657,
+  "<|box_end|>": 151649,
+  "<|box_start|>": 151648,
+  "<|endoftext|>": 151643,
+  "<|file_sep|>": 151664,
+  "<|fim_middle|>": 151660,
+  "<|fim_pad|>": 151662,
+  "<|fim_prefix|>": 151659,
+  "<|fim_suffix|>": 151661,
+  "<|im_end|>": 151645,
+  "<|im_start|>": 151644,
+  "<|image_pad|>": 151655,
+  "<|object_ref_end|>": 151647,
+  "<|object_ref_start|>": 151646,
+  "<|quad_end|>": 151651,
+  "<|quad_start|>": 151650,
+  "<|repo_name|>": 151663,
+  "<|video_pad|>": 151656,
+  "<|vision_end|>": 151653,
+  "<|vision_pad|>": 151654,
+  "<|vision_start|>": 151652
+}
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -0,0 +1,54 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- messages[0]['content'] }}
+    {%- else %}
+        {{- 'You are Qwen, created by Alibaba Cloud. You are a helpful assistant.' }}
+    {%- endif %}
+    {{- "\n\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0]['role'] == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0]['content'] + '<|im_end|>\n' }}
+    {%- else %}
+        {{- '<|im_start|>system\nYou are Qwen, created by Alibaba Cloud. You are a helpful assistant.<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) or (message.role == "assistant" and not message.tool_calls) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {{- '<|im_start|>' + message.role }}
+        {%- if message.content %}
+            {{- '\n' + message.content }}
+        {%- endif %}
+        {%- for tool_call in message.tool_calls %}
+            {%- if tool_call.function is defined %}
+                {%- set tool_call = tool_call.function %}
+            {%- endif %}
+            {{- '\n<tool_call>\n{"name": "' }}
+            {{- tool_call.name }}
+            {{- '", "arguments": ' }}
+            {{- tool_call.arguments | tojson }}
+            {{- '}\n</tool_call>' }}
+        {%- endfor %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if (loop.index0 == 0) or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+{%- endif %}
--- a/config.json
+++ b/config.json
@@ -0,0 +1,58 @@
+{
+  "architectures": [
+    "Qwen2ForCausalLM"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "dtype": "bfloat16",
+  "eos_token_id": 151645,
+  "hidden_act": "silu",
+  "hidden_size": 1536,
+  "initializer_range": 0.02,
+  "intermediate_size": 8960,
+  "layer_types": [
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 32768,
+  "max_window_layers": 21,
+  "model_type": "qwen2",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 2,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "transformers_version": "4.57.6",
+  "use_cache": true,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}
--- a/configuration.json
+++ b/configuration.json
@@ -0,0 +1 @@
+{}
--- a/generation_config.json
+++ b/generation_config.json
@@ -0,0 +1,14 @@
+{
+  "bos_token_id": 151643,
+  "pad_token_id": 151643,
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "repetition_penalty": 1.1,
+  "temperature": 0.7,
+  "top_p": 0.8,
+  "top_k": 20,
+  "transformers_version": "4.37.0"
+}
--- a/merges.txt
+++ b/merges.txt
--- a/model-00001-of-00057.safetensors
+++ b/model-00001-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40b7e6a365b661ef9a18e28c26e71c942a28c7eda6ded217cc6641333e026a73
+size 466747528
--- a/model-00002-of-00057.safetensors
+++ b/model-00002-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0e424dec445006aa68ba120d04ff5fde7137ffd7ebfaa694ef9a79aba9a7d0be
+size 38540176
--- a/model-00003-of-00057.safetensors
+++ b/model-00003-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:107f4d225c81534f5ca05c78d577ecf0de1532bcb0c1a68c83365f82efed1794
+size 61353056
--- a/model-00004-of-00057.safetensors
+++ b/model-00004-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e72c454caa0c0f9c5241a00a7bdb29a28d39cb1cb46105ea5e80e9935d48099
+size 59769200
--- a/model-00005-of-00057.safetensors
+++ b/model-00005-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b422b8db4bb1e9b513924ca9e8970baf1fb0e2d28a07a143116aec50ebf9e0c
+size 38546536
--- a/model-00006-of-00057.safetensors
+++ b/model-00006-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27c8e431125b19bf2193c62e9acf40b2003ea674a891442d83f6c701e92cd96c
+size 55050496
--- a/model-00007-of-00057.safetensors
+++ b/model-00007-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43a1bb4fddd96c3936958f20f900e9374f7216bd7f0f1a138d94fc91840494a0
+size 38546536
--- a/model-00008-of-00057.safetensors
+++ b/model-00008-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de579e160842610e4bf68ddf214d9cbb16ceb7964c01559fdd2abb09321cf387
+size 55050496
--- a/model-00009-of-00057.safetensors
+++ b/model-00009-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5502ae20af0b18fb010125ef28364ec6479adc747e24b9e9139c6f8f3952d0f
+size 38546536
--- a/model-00010-of-00057.safetensors
+++ b/model-00010-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:960196906e2f08ed4047e0373c36679d4693f8e6fabbef4c9b0115d27be0400f
+size 55050496
--- a/model-00011-of-00057.safetensors
+++ b/model-00011-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5511820e9e5f5e2f22336d11034e74a90aeae4a5c9fd45dab1d742f991bc59b
+size 38546536
--- a/model-00012-of-00057.safetensors
+++ b/model-00012-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d626fb78d605e8ded76ebed21c5d56093444039e4f033e4be8c5792da388003
+size 55050496
--- a/model-00013-of-00057.safetensors
+++ b/model-00013-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa0597fb6748edd075389a3bf03d5ff26e45277e46c589b26b4169a3e1e6c300
+size 38546536
--- a/model-00014-of-00057.safetensors
+++ b/model-00014-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc5713f7f58e4471dc6024d398fd583ab87ccd4d924073dae498cd290afbcb78
+size 55050496
--- a/model-00015-of-00057.safetensors
+++ b/model-00015-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bf78a5882022a5690824dee00f2b622057d62d88b9edb5a7bb158689eff9ee0
+size 38546536
--- a/model-00016-of-00057.safetensors
+++ b/model-00016-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94d03f1a2bf7d4432a16aaa42be2ea0c133b49092d768a7142a927761eb47937
+size 55050496
--- a/model-00017-of-00057.safetensors
+++ b/model-00017-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51b819f0ecf1e11c8615723717faf65bc97bf62fd5f6dcde0223675b9d47a268
+size 38546536
--- a/model-00018-of-00057.safetensors
+++ b/model-00018-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dba6f1d847b5c9b6fb81504a8e69e98677013835110ecfeb11ec0e4cdbf3bfb2
+size 55050496
--- a/model-00019-of-00057.safetensors
+++ b/model-00019-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7214926cb491e9ce4fdb493e97210531d7e52e9c9a2beedecbab2bf2cd2f1f12
+size 38546536
--- a/model-00020-of-00057.safetensors
+++ b/model-00020-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b74b2b946b5809eb8ad4a4797d9bf5f2b50b7867d1c5a7deaab410e2cd983be8
+size 55050496
--- a/model-00021-of-00057.safetensors
+++ b/model-00021-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:260746ed008a026b63945f0883b8cceb1f1c0d8f1a05ed47be4fe7af36061d7b
+size 38546536
--- a/model-00022-of-00057.safetensors
+++ b/model-00022-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e19822c7fc679bb79f53353b1136748d6b4ad7126393a1adbde7b1012764fcb1
+size 55050496
--- a/model-00023-of-00057.safetensors
+++ b/model-00023-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cdbb586278e92f657d33c1a202dcf4dd6b5e258db6437bdfbea6ac38289e4be
+size 38546544
--- a/model-00024-of-00057.safetensors
+++ b/model-00024-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29cf52078e1a6ef26b017f6ea0f410b38973dd1ff6be5870c8ad807a137c1af7
+size 55050496
--- a/model-00025-of-00057.safetensors
+++ b/model-00025-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9057960ad72e5b8f6c417500efe829840b1d209953c7588c0c01697d486e38d
+size 38546544
--- a/model-00026-of-00057.safetensors
+++ b/model-00026-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26cedc91f5b3e8219448f9b34ccfd56a9f8aeb0616582d5d8837d5d8086330fc
+size 55050496
--- a/model-00027-of-00057.safetensors
+++ b/model-00027-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b71d9ad7856d40b601f5a03860545f2b43df13c52e154ca41491cf08f2777af0
+size 38546544
--- a/model-00028-of-00057.safetensors
+++ b/model-00028-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8f712681a834c3f680acbe11f7b35dce89bd3763c486a71f28bdabc0345692e
+size 55050496
--- a/model-00029-of-00057.safetensors
+++ b/model-00029-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc08b6f95bd1235edf0d2e1ee8e8c787133906494d68127b0a6256286696ee4f
+size 38546544
--- a/model-00030-of-00057.safetensors
+++ b/model-00030-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4436dcae1abb5721cd1740583c8fdd5d6917577e6e57c6a61f22e361a6031a7d
+size 55050496
--- a/model-00031-of-00057.safetensors
+++ b/model-00031-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b5165e80fc0fb8fd8e781a33e586535281cdba313766326b2c8a01eca1b51cd
+size 38546544
--- a/model-00032-of-00057.safetensors
+++ b/model-00032-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c4c097b5664f5e4f0fa6240ad22a0e728a92dfffc9e4e7f8c438d6f4f7088de
+size 55050496
--- a/model-00033-of-00057.safetensors
+++ b/model-00033-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:507e36dd372754a19cf75c14baecaa44594b3a2090577ab31aaf9d86b289e6e0
+size 38546544
--- a/model-00034-of-00057.safetensors
+++ b/model-00034-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4623bf9fc7f2bcf68cdf5bf3f88ee61505a41fcc74e28c90d448b3f6ccdefc8
+size 55050496
--- a/model-00035-of-00057.safetensors
+++ b/model-00035-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50a2fc54c87bc5f00275c6aeabe443e32295f95c1fe638968be0f4015543d059
+size 38546544
--- a/model-00036-of-00057.safetensors
+++ b/model-00036-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cd4da12a345f292720ac49729277613070051715201ff60001b6069995ea7e2
+size 55050496
--- a/model-00037-of-00057.safetensors
+++ b/model-00037-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fadf5efa6c63de94ac68823a8d8b94eb37063d3fd92058271f55bbef41bc5952
+size 38546544
--- a/model-00038-of-00057.safetensors
+++ b/model-00038-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd4881fe3cd500ed5cf7f4d75c6986430586f107daa796072721a2491aadd0a4
+size 55050496
--- a/model-00039-of-00057.safetensors
+++ b/model-00039-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:478372df4b06c239a57eb2fd0a56703a6cebf5da36a7ea86ee6360b0450bc22f
+size 38546544
--- a/model-00040-of-00057.safetensors
+++ b/model-00040-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ea8051e5ff9bd279e92ad1506d866c2ecbdd945e3700cfd793c9caf7c9fda24
+size 55050496
--- a/model-00041-of-00057.safetensors
+++ b/model-00041-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17b9e9feb1763522dd2e6cdb7d494e1718af1100ac64b306f6936c31d89217ae
+size 38546544
--- a/model-00042-of-00057.safetensors
+++ b/model-00042-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:575563b5969455602b208be41fbc8bedd27f5a0fdf8ae5cfc156ec557b09a0d7
+size 55050496
--- a/model-00043-of-00057.safetensors
+++ b/model-00043-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:651673ab04ecae55e7c558c99c5e5c5a9f09b4744695095b8cd1c911c659c8db
+size 38546544
--- a/model-00044-of-00057.safetensors
+++ b/model-00044-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:70f4830f07fe2791b5e1f62ae129b8d7ec2d9fd77ed212281d529a3cedb4e237
+size 55050496
--- a/model-00045-of-00057.safetensors
+++ b/model-00045-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4098ae1e96480159035b629ffb7b80b31fff216c1ca966280d7d673e5672d8fa
+size 38546544
--- a/model-00046-of-00057.safetensors
+++ b/model-00046-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ef1c309e528f7a231880fed9558834f8cf0404084712c892551cc8168a884aa
+size 55050496
--- a/model-00047-of-00057.safetensors
+++ b/model-00047-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e992519bcb9cadbd852193798259ad9710acb7d16e1afeec139ff1572f06fef
+size 38546544
--- a/model-00048-of-00057.safetensors
+++ b/model-00048-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e937439fe4c443b25d753422fa7ac838aaa34ca5ce0952af2d17e8b5ac014b8
+size 55050496
--- a/model-00049-of-00057.safetensors
+++ b/model-00049-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc34d6d302f7ed475e4937054136c72ceace861b3a309b4d1eaded8693fc1fec
+size 38546544
--- a/model-00050-of-00057.safetensors
+++ b/model-00050-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b86abe7ee395a1dbbf16dc34ce2e217daedb9c5d406667f154937b0d0dcceea6
+size 55050496
--- a/model-00051-of-00057.safetensors
+++ b/model-00051-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:089b7b9ba149df121622a7a87fcebef93658960f67aa398974382e09bf0eec89
+size 38546544
--- a/model-00052-of-00057.safetensors
+++ b/model-00052-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71c1362e4063164535d363adb831f9bfe5dc69016ebdc0115b6c152decd912ec
+size 55050496
--- a/model-00053-of-00057.safetensors
+++ b/model-00053-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f7172e683cd9661414244c6e12bdef297373966e152345ca360cbdcc5f939e6
+size 38546544
--- a/model-00054-of-00057.safetensors
+++ b/model-00054-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fd66b78de9e51288dc84f6d4ea435586bfc6dbc17f0393d61043bbf4cb7ae71
+size 55050496
--- a/model-00055-of-00057.safetensors
+++ b/model-00055-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10d83ed67dafeeb78c5508b58c6ed34a0557257402d87957ffe7dcec949aa17a
+size 38546544
--- a/model-00056-of-00057.safetensors
+++ b/model-00056-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17cb8f0e6a2f4914e5fced6e69b6a65cda80d077e754754e6e9d63603ed3caec
+size 55050496
--- a/model-00057-of-00057.safetensors
+++ b/model-00057-of-00057.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aa220232e481fd80a3472c157de1bd936cb479716e2fd8f9a42c50b7da90e49
+size 27534784
--- a/model.safetensors.index.json
+++ b/model.safetensors.index.json
@@ -0,0 +1,346 @@
+{
+  "metadata": {
+    "total_parameters": 1543714304,
+    "total_size": 3087428608
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00057.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00003-of-00057.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00003-of-00057.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00002-of-00057.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00003-of-00057.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00003-of-00057.safetensors",
+    "model.layers.0.self_attn.k_proj.bias": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.q_proj.bias": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.v_proj.bias": "model-00002-of-00057.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00002-of-00057.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00005-of-00057.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00005-of-00057.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00004-of-00057.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00004-of-00057.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00005-of-00057.safetensors",
+    "model.layers.1.self_attn.k_proj.bias": "model-00003-of-00057.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00003-of-00057.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00004-of-00057.safetensors",
+    "model.layers.1.self_attn.q_proj.bias": "model-00003-of-00057.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00003-of-00057.safetensors",
+    "model.layers.1.self_attn.v_proj.bias": "model-00003-of-00057.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00003-of-00057.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00023-of-00057.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00023-of-00057.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00022-of-00057.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00022-of-00057.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00023-of-00057.safetensors",
+    "model.layers.10.self_attn.k_proj.bias": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.q_proj.bias": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.v_proj.bias": "model-00021-of-00057.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00021-of-00057.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00025-of-00057.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00025-of-00057.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00024-of-00057.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00024-of-00057.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00025-of-00057.safetensors",
+    "model.layers.11.self_attn.k_proj.bias": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.q_proj.bias": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.v_proj.bias": "model-00023-of-00057.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00023-of-00057.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00027-of-00057.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00027-of-00057.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00026-of-00057.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00026-of-00057.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00027-of-00057.safetensors",
+    "model.layers.12.self_attn.k_proj.bias": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.q_proj.bias": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.v_proj.bias": "model-00025-of-00057.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00025-of-00057.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00029-of-00057.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00029-of-00057.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00028-of-00057.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00028-of-00057.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00029-of-00057.safetensors",
+    "model.layers.13.self_attn.k_proj.bias": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.q_proj.bias": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.v_proj.bias": "model-00027-of-00057.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00027-of-00057.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00031-of-00057.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00031-of-00057.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00030-of-00057.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00030-of-00057.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00031-of-00057.safetensors",
+    "model.layers.14.self_attn.k_proj.bias": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.q_proj.bias": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.v_proj.bias": "model-00029-of-00057.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00029-of-00057.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00033-of-00057.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00033-of-00057.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00032-of-00057.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00032-of-00057.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00033-of-00057.safetensors",
+    "model.layers.15.self_attn.k_proj.bias": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.q_proj.bias": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.v_proj.bias": "model-00031-of-00057.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00031-of-00057.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00035-of-00057.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00035-of-00057.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00034-of-00057.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00034-of-00057.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00035-of-00057.safetensors",
+    "model.layers.16.self_attn.k_proj.bias": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.q_proj.bias": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.v_proj.bias": "model-00033-of-00057.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00033-of-00057.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00037-of-00057.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00037-of-00057.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00036-of-00057.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00036-of-00057.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00037-of-00057.safetensors",
+    "model.layers.17.self_attn.k_proj.bias": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.q_proj.bias": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.v_proj.bias": "model-00035-of-00057.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00035-of-00057.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00039-of-00057.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00039-of-00057.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00038-of-00057.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00038-of-00057.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00039-of-00057.safetensors",
+    "model.layers.18.self_attn.k_proj.bias": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.q_proj.bias": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.v_proj.bias": "model-00037-of-00057.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00037-of-00057.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00041-of-00057.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00041-of-00057.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00040-of-00057.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00040-of-00057.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00041-of-00057.safetensors",
+    "model.layers.19.self_attn.k_proj.bias": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.q_proj.bias": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.v_proj.bias": "model-00039-of-00057.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00039-of-00057.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00007-of-00057.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00007-of-00057.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00006-of-00057.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00006-of-00057.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00007-of-00057.safetensors",
+    "model.layers.2.self_attn.k_proj.bias": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.q_proj.bias": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.v_proj.bias": "model-00005-of-00057.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00005-of-00057.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00043-of-00057.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00043-of-00057.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00042-of-00057.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00042-of-00057.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00043-of-00057.safetensors",
+    "model.layers.20.self_attn.k_proj.bias": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.q_proj.bias": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.v_proj.bias": "model-00041-of-00057.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00041-of-00057.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00045-of-00057.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00045-of-00057.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00044-of-00057.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00044-of-00057.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00045-of-00057.safetensors",
+    "model.layers.21.self_attn.k_proj.bias": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.q_proj.bias": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.v_proj.bias": "model-00043-of-00057.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00043-of-00057.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00047-of-00057.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00047-of-00057.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00046-of-00057.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00046-of-00057.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00047-of-00057.safetensors",
+    "model.layers.22.self_attn.k_proj.bias": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.q_proj.bias": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.v_proj.bias": "model-00045-of-00057.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00045-of-00057.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00049-of-00057.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00049-of-00057.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00048-of-00057.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00048-of-00057.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00049-of-00057.safetensors",
+    "model.layers.23.self_attn.k_proj.bias": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.q_proj.bias": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.v_proj.bias": "model-00047-of-00057.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00047-of-00057.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00051-of-00057.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00051-of-00057.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00050-of-00057.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00050-of-00057.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00051-of-00057.safetensors",
+    "model.layers.24.self_attn.k_proj.bias": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.q_proj.bias": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.v_proj.bias": "model-00049-of-00057.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00049-of-00057.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00053-of-00057.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00053-of-00057.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00052-of-00057.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00052-of-00057.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00053-of-00057.safetensors",
+    "model.layers.25.self_attn.k_proj.bias": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.q_proj.bias": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.v_proj.bias": "model-00051-of-00057.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00051-of-00057.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00055-of-00057.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00055-of-00057.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00054-of-00057.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00054-of-00057.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00055-of-00057.safetensors",
+    "model.layers.26.self_attn.k_proj.bias": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.q_proj.bias": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.v_proj.bias": "model-00053-of-00057.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00053-of-00057.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00057-of-00057.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00057-of-00057.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00056-of-00057.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00056-of-00057.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00057-of-00057.safetensors",
+    "model.layers.27.self_attn.k_proj.bias": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.q_proj.bias": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.v_proj.bias": "model-00055-of-00057.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00055-of-00057.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00009-of-00057.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00009-of-00057.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00008-of-00057.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00008-of-00057.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00009-of-00057.safetensors",
+    "model.layers.3.self_attn.k_proj.bias": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.q_proj.bias": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.v_proj.bias": "model-00007-of-00057.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00007-of-00057.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00011-of-00057.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00011-of-00057.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00010-of-00057.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00010-of-00057.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00011-of-00057.safetensors",
+    "model.layers.4.self_attn.k_proj.bias": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.q_proj.bias": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.v_proj.bias": "model-00009-of-00057.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00009-of-00057.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00013-of-00057.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00013-of-00057.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00012-of-00057.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00012-of-00057.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00013-of-00057.safetensors",
+    "model.layers.5.self_attn.k_proj.bias": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.q_proj.bias": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.v_proj.bias": "model-00011-of-00057.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00011-of-00057.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00015-of-00057.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00015-of-00057.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00014-of-00057.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00014-of-00057.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00015-of-00057.safetensors",
+    "model.layers.6.self_attn.k_proj.bias": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.q_proj.bias": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.v_proj.bias": "model-00013-of-00057.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00013-of-00057.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00017-of-00057.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00017-of-00057.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00016-of-00057.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00016-of-00057.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00017-of-00057.safetensors",
+    "model.layers.7.self_attn.k_proj.bias": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.q_proj.bias": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.v_proj.bias": "model-00015-of-00057.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00015-of-00057.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00019-of-00057.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00019-of-00057.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00018-of-00057.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00018-of-00057.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00019-of-00057.safetensors",
+    "model.layers.8.self_attn.k_proj.bias": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.q_proj.bias": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.v_proj.bias": "model-00017-of-00057.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00017-of-00057.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00021-of-00057.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00021-of-00057.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00020-of-00057.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00020-of-00057.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00021-of-00057.safetensors",
+    "model.layers.9.self_attn.k_proj.bias": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.q_proj.bias": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.v_proj.bias": "model-00019-of-00057.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00019-of-00057.safetensors",
+    "model.norm.weight": "model-00057-of-00057.safetensors"
+  }
+}
--- a/special_tokens_map.json
+++ b/special_tokens_map.json
@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
--- a/tokenizer.json
+++ b/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9c5ae00e602b8860cbd784ba82a8aa14e8feecec692e7076590d014d7b7fdafa
+size 11421896
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@@ -0,0 +1,207 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
--- a/vocab.json
+++ b/vocab.json