Initialize the repository

.gitattributes

@@ -33,3 +33,8 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.axmodel filter=lfs diff=lfs merge=lfs -text
+*.jpg filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text
+*.mp4 filter=lfs diff=lfs merge=lfs -text
+examples/red-panda.mp4 filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,4 @@
+__pycache__
+build-output/
+tmp/
+*.safetensors

README.md

@@ -1,3 +1,47 @@
 ---
 license: mit
 ---
+
+# RAG.AXERA DEMO
+
+![rag_demo](assets/demo.png)
+
+## 项目说明
+
+```sh
+(hf) ➜  rag.axera git:(main) ✗ tree -L 2
+.
+├── assets
+│   └── demo.png
+├── config.py     # 配置 axmodel, tokenizer 文件路径
+├── data
+├── gui.py        # RAG 交互式程序 
+├── index         # 文档编码向量索引保存位置
+│   ├── docs.index
+│   └── docs.pkl
+├── llm_api.py    # llm 主程序 
+├── models        # axmodel 模型存储位置
+│   ├── Qwen2.5-1.5B-Instruct_axmodel
+│   └── Qwen3-Embedding-0.6B_axmodel
+├── pdf_sample    # 示例 pdf 文件
+│   └── introduction.pdf
+├── rag_engine.py # 文档向量编码程序 
+├── README.md
+├── requirements.txt
+├── tokenizer
+│   ├── Qwen2.5-1.5B-Instruct
+│   └── Qwen3-Embedding-0.6B
+└── utils
+    └── infer_func.py
+
+11 directories, 11 files
+```
+
+## 运行
+
+在 `AXCL` 机器或 `AX650` 开发板上启动两个终端界面, 分别运行下面的命令:
+
+```sh
+python3 llm_api.py # 在 AX650 或 AXCL 开发板启动 llm 服务
+python3 gui.py     # 启动交互式界面
+```

config.py

@@ -0,0 +1,23 @@
+# config.py
+import os
+from dotenv import load_dotenv
+
+load_dotenv()
+
+# models and paths
+LLM_HF_MODEL = os.getenv("LLM_HF_MODEL", "./tokenizer/Qwen2.5-1.5B-Instruct") # 只需要根据实际路径修改这里的模型路径即可
+LLM_AX_MODEL = os.getenv("LLM_AX_MODEL", "./models/Qwen2.5-1.5B-Instruct_axmodel")
+EMBED_HF_MODEL = os.getenv("EMBED_HF_MODEL", "./tokenizer/Qwen3-Embedding-0.6B")
+EMBED_AX_MODEL = os.getenv("EMBED_AX_MODEL", "./models/Qwen3-Embedding-0.6B_axmodel")
+
+# API URL
+LLM_API_PORT = int(os.getenv("LLM_API_PORT", "8000"))
+LLM_API_URL = os.getenv("LLM_API_URL", f"http://127.0.0.1:{LLM_API_PORT}/generate") # 具体函数名字需要对应修改
+PORT = int(os.getenv("PORT", "7860"))
+
+# Index paths
+INDEX_DIR = os.getenv("INDEX_DIR", "index")
+INDEX_FILE = os.path.join(INDEX_DIR, "docs.index")
+EMBEDDINGS_FILE = os.path.join(INDEX_DIR, "docs.pkl")
+
+os.makedirs(INDEX_DIR, exist_ok=True)

gui.py

@@ -0,0 +1,65 @@
+# gui.py
+import gradio as gr
+from rag_engine import ask_question, build_index, stream_answer
+from config import PORT
+import time
+
+chat_history = []
+
+def handle_upload(file):
+    if file is None:
+        return "❌ 请上传文件", ""
+    try:
+        result = build_index(file.name)  # file.name 是本地路径
+        return result, ""
+    except Exception as e:
+        return f"❌ 构建索引失败：{str(e)}", ""
+
+def handle_chat(message, history):
+    history = history or []
+    if not message.strip():
+        return "", history
+    try:
+        # 使用流式响应
+        history.append((message, ""))
+        full_response = ""
+
+        # 获取流式响应生成器
+        answer_generator = stream_answer(message)
+
+        # 逐个token添加到聊天历史
+        for token in answer_generator:
+            full_response += token
+            history[-1] = (message, full_response)
+            yield "", history
+            time.sleep(0.02)  # 添加微小延迟使输出更平滑
+
+        # 流结束后添加一点停顿
+        time.sleep(0.1)
+        yield "", history
+
+    except Exception as e:
+        history.append((message, f"⚠️ 出错了：{str(e)}"))
+        return "", history
+
+with gr.Blocks(title="RAG 文档问答系统") as demo:
+    gr.Markdown("## 🤖 AXERA RAG 文档问答\n请上传 PDF 或 TXT 文件并提问")
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            file_input = gr.File(label="📄 上传文件", file_types=[".pdf", ".txt"])
+            upload_btn = gr.Button("📥 上传并构建索引")
+            upload_status = gr.Textbox(label="", interactive=False)
+
+        with gr.Column(scale=2):
+            chatbot = gr.Chatbot(height=400, label="🧠 问答对话")
+            with gr.Row():
+                message = gr.Textbox(placeholder="请输入你的问题，按 Shift + Enter 发送", show_label=False, lines=2)
+                send_btn = gr.Button("🚀 发送")
+
+    upload_btn.click(fn=handle_upload, inputs=[file_input], outputs=[upload_status, message])
+    send_btn.click(fn=handle_chat, inputs=[message, chatbot], outputs=[message, chatbot])
+    message.submit(fn=handle_chat, inputs=[message, chatbot], outputs=[message, chatbot])
+
+# 启用队列并启动
+demo.queue().launch(server_port=PORT)

index/docs.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad92dcae43938a9ca8afbfff7c9cf6c671a2290e4839f0199a6f834eefdfceac
+size 5705

llm_api.py

@@ -0,0 +1,142 @@
+# llm_api.py
+from fastapi import FastAPI, HTTPException
+from fastapi.responses import StreamingResponse
+from pydantic import BaseModel
+from typing import Optional
+import uvicorn
+import numpy as np
+import os
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig
+from config import LLM_HF_MODEL, LLM_AX_MODEL, LLM_API_PORT
+
+
+app = FastAPI(title="Fast-API", description="本地推理接口")
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+"""
+axengine 相关
+"""
+from ml_dtypes import bfloat16
+from utils.infer_func import InferManager
+
+# 定义全局变量，但先不初始化
+tokenizer = None
+imer = None
+embeds = None
+
+def init_model():
+    global tokenizer, imer, embeds
+    if tokenizer is None:  # 防止重复初始化
+        cfg = AutoConfig.from_pretrained(LLM_HF_MODEL) 
+        imer = InferManager(cfg, LLM_AX_MODEL, model_type="qwen2")
+        embeds = np.load(os.path.join(LLM_AX_MODEL, "model.embed_tokens.weight.npy"))
+        # 加载 tokenizer
+        tokenizer = AutoTokenizer.from_pretrained(LLM_HF_MODEL, trust_remote_code=True)
+        print("✅ 模型加载完成。")
+
+# 添加 FastAPI 的启动事件
+@app.on_event("startup")
+async def startup_event():
+    init_model()
+
+class GenRequest(BaseModel):
+    prompt: str
+    max_tokens: Optional[int] = 1024
+    temperature: Optional[float] = 0.6
+    top_p: Optional[float] = 0.9
+
+class GenResponse(BaseModel):
+    text: str
+
+@app.post("/generate", response_model=GenResponse)
+def generate_text(req: GenRequest):
+    try:
+        # input_ids = tokenizer(req.prompt, return_tensors="pt").input_ids.to(device)
+
+        # with torch.no_grad():
+        #     output_ids = model.generate(
+        #         input_ids=input_ids,
+        #         max_new_tokens=req.max_tokens,
+        #         temperature=req.temperature,
+        #         top_p=req.top_p,
+        #         # do_sample=True,
+        #         eos_token_id=tokenizer.eos_token_id
+        #     )
+
+        # response_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)
+        messages = [
+            {"role": "system", "content": "你的名字叫做 [AXERA-RAG 助手]. 你是一个高效、精准的问答助手. 你可以根据上下文内容, 回答用户提出的问题, 回答时不要提及多余的、无用的内容, 且仅输出你的回答."},
+            {"role": "user", "content": req.prompt}
+        ]
+        text = tokenizer.apply_chat_template(
+            messages,
+            tokenize=False,
+            add_generation_prompt=True
+        )
+        model_inputs = tokenizer([text], return_tensors="pt").to(device)
+
+        """
+        axengine 框架模型推理
+        """
+        input_ids = model_inputs['input_ids']
+        inputs_embeds = np.take(embeds, input_ids.cpu().numpy(), axis=0)
+        prefill_data = inputs_embeds
+        prefill_data = prefill_data.astype(bfloat16)
+        token_ids = input_ids[0].cpu().numpy().tolist()
+        generated_text = ""
+
+        def generate_stream():
+            nonlocal token_ids, generated_text
+            token_ids = imer.prefill(tokenizer, token_ids, prefill_data[0], slice_len=128)
+            generated_text += tokenizer.decode(token_ids[-1], skip_special_tokens=True)
+
+            # response_text = imer.decode(tokenizer, token_ids, embeds, slice_len=128)
+            # 去掉 prompt 的前缀, 只保留生成部分
+            # generated_text = response_text[len(req.prompt):].strip()
+            # generated_text = response_text
+            # return GenResponse(text=generated_text)
+
+            # 流式输出控制
+            prefill_word = tokenizer.decode(token_ids[-1], skip_special_tokens=True)
+            prefill_word = prefill_word.strip().replace("\n", "\\n").replace("\"", "\\\"")
+
+            seq_len = len(token_ids) - 1
+            prefill_len = 128
+            for step_idx in range(imer.max_seq_len):
+                if prefill_len > 0 and step_idx < seq_len:
+                    continue
+                token_ids, next_token_id = imer.decode_next_token(tokenizer, token_ids, embeds, slice_len=128, step_idx=step_idx)
+                if next_token_id == tokenizer.eos_token_id and next_token_id > seq_len:
+                    break
+                try:
+                    if next_token_id is not None:
+                        word = tokenizer.decode([next_token_id], skip_special_tokens=True)
+                        generated_text += word
+                        if prefill_word is not None:
+                            word = prefill_word + word
+                            prefill_word = None
+                        # 以适合前端处理的 SSE 格式输出
+                        # 处理特殊字符
+                        word = word.strip().replace("\n", "\\n").replace("\"", "\\\"")
+                        # import pdb; pdb.set_trace()
+                        yield f"data: {{\"token\": \"{word}\"}}\n\n"
+                except Exception as e:
+                    print(f"Error decoding token {next_token_id}: {e}")
+
+        return StreamingResponse(
+            generate_stream(),
+            media_type="text/event-stream",  # 必须使用SSE格式
+            headers={
+                "Cache-Control": "no-cache",
+                "Connection": "keep-alive",
+                "X-Accel-Buffering": "no"  # 禁用Nginx缓冲
+            }
+        )
+
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=LLM_API_PORT, reload=False)

models/Qwen2.5-1.5B-Instruct_axmodel/model.embed_tokens.weight.bfloat16.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b2c594e5f910978ef413824340261a6055c5bb905dcfefceed9d30dd2b80637e
+size 466747392

models/Qwen2.5-1.5B-Instruct_axmodel/model.embed_tokens.weight.float32.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2517b46ffa436067fc5e2bd7e191107c2c5c9f29892358e8639957b0057a287
+size 933494784

models/Qwen2.5-1.5B-Instruct_axmodel/model.embed_tokens.weight.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fcec2b0923f51df5891df10e715e3204e3bb039f26780112ade2f1b7da997bef
+size 933494912

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l0_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63ef37adbe925889dc0be44e9ba35aefec5eb8f1824f3c785510765f1a1bc6fe
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l10_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe006a7dd1096a5d8c68ef9d70ed49f9f1ed4769f122080a735da8d04f025e02
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l11_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e350949ca5683b3084ae4241bdd1fddcaebe5ef8e70c0d8aef339a3b5e693c2
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l12_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39501bd3523287d5fa67973abf11a91f4a67c79ae096c0d31c7e07fe01427690
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l13_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:627aa4354f2c7327bf2c053d2a5f6acecde55a406ebd0713cd7c7feffa6855c0
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l14_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8d04ec0cb7ba6170c18c6fb6ccdec8586d3568717d93de562ac31b032fd34d5
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l15_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6d0ab59dc0bf214354b6ed6b9b155d5ff4fb3e7cc10975cc0c1652e39a660f7a
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l16_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cd4e7800e81bbed49d15c4354ad763a699af9f8ec3413dd9d1cc9a6b49d9c6fb
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l17_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b8d5ec8f6dafda2b90ede8a630a98e865c3b6012995c17e6a6b5b606f269237
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l18_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79b37891f1710e559ba39604b83badad68c4ec971fb6883733e963afff250c1b
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l19_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3863cbeb6191f43dfc7e3d31c3bfe4bf4dccba3ea9769003c52754ec2fb4be12
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l1_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4755cfff48189833c5062bc07683e551443424b985ed1575b8db4a2297ca0894
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l20_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f34c1d1589a2f2545f73eb03dc231e2fd434e26fb7015b0ae4e78cb9ab6329c2
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l21_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64df1cce0340b34a586363b5d86158289501895836667e084ca172c81ae5dbcf
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l22_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c60464b0a790a9e39934a35f7bdc4b1948903cb74c93c3321313a4f9d2e9b6b
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l23_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a71d4d9806fc91f4eeddc7c336074f26fc10a436163537c1f6fb5f92ab0c13c2
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l24_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:678177517e5efaa54688a2debe1402c83a056285062c00d119ec0f53f80b22ff
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l25_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ea4077e32a81594ed2eab85edca2579aaa07d6610c38bae722d50519db54f7c5
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l26_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e23d6bb592bb43cc68f2cee5d95be7f97401be13f8bb882bfe5ca23d5f7a5a1
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l27_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cd266fdd498697b600309ffd89c3f0e43d8dbb7aaf3f4f0413520c9500e7ab6
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l2_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:beae1c79a0455b1d66747e0078a2057d299360558a4a4f7c81ea4a9062a8fa1b
+size 67102542

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l3_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:716ea68fe92de87bb162f8033bfbe57448b4eca36ce25af0c5b74078240190ee
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l4_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47d97c2421ccbb5c64e50337a1f7b4615f5f51ad2c50a16f2f9afef951e33056
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l5_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eda5ef5b2eae8c023fa47f13ffc1757c921b0cecfcfee2d3dedf9f3ea8079d73
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l6_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:693e656bfc8c54beb76a194be5839ffe6535691b5fc3c74e6a7ad8ea291e9c89
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l7_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d842213daf2da92c1f4df7d981233f5faa64344f4ddec8009c9874a54fa650f
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l8_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:957387916555920f33d3207db940e1918f64c7c389c31f808d324d2e3d9d0d95
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_p128_l9_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3da1f60bc8b971ef8ef6098cf11e4b44cbc9c819cc6cfeb3b646941e50d13a9
+size 67100526

models/Qwen2.5-1.5B-Instruct_axmodel/qwen2_post.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1d25088c38ccf35bd82c7d871878ab1cbe8512e00130ca871eb5e9601768a94
+size 254449571

models/Qwen3-Embedding-0.6B_axmodel/model.embed_tokens.weight.bfloat16.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a55b140d86852835bd18d8200222a9f302340730f0670eb7e23a4895e5489033
+size 310618112

models/Qwen3-Embedding-0.6B_axmodel/model.embed_tokens.weight.float32.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7a027c062fb61cd505e046bc832345be155e1eb2fab629675cebe7973646c85
+size 621236224

models/Qwen3-Embedding-0.6B_axmodel/model.embed_tokens.weight.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bbdc47aee1b4cdb97a42a255306d4e0a1cb52f797bfdc32f94469eb0cd0744e
+size 621236352

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l0_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:025e17a92f3f19d58a36ef119294598073c4ccdc794aa9d4a2845a99b0c6b53d
+size 28019747

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l10_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:95d809d6cc6889517b1aad7a4e62e51ffbf75580dda5ceafb667dbd5ac10ba6e
+size 28019779

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l11_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7fb730805877eceea4aa037694bc2abb830fa960c6666d24b976d7ae35c058d0
+size 28018723

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l12_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eef7266dff6af522a0a63095067a1c7823a9a1213e7bd498bcdb97f2814523ba
+size 28019427

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l13_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9322be28dad0729b006238364297659594e2193a97a37556fb06f63d3fec9fa0
+size 28019459

models/Qwen3-Embedding-0.6B_axmodel/qwen3_p128_l14_together.axmodel

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c5a8f239e4a4e793e0bdf86226c08c2089f5199118bd38d2be4957f9b7023dda
+size 28018723