初始化项目，由ModelHub XC社区提供模型

Model: Smith-3/simon-fcyt-umss-v2 Source: Original Platform
2026-06-19 04:51:18 +08:00
commit cd15bc0024
16 changed files with 328368 additions and 0 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,41 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt 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
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz 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
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl 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
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* 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
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz 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
+LFM2.5-350M.Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
+LFM2.5-350M.Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
+LFM2.5-350M.Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
+LFM2.5-350M.Q6_K.gguf filter=lfs diff=lfs merge=lfs -text
+LFM2.5-350M.Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
+LFM2.5-350M.F16.gguf filter=lfs diff=lfs merge=lfs -text
--- a/LFM2.5-350M.F16.gguf
+++ b/LFM2.5-350M.F16.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:921811fe3655a5f997c99d33c7d9db6684a73fd5728e5a8bad77c7f766922f60
+size 711483968
--- a/LFM2.5-350M.Q3_K_M.gguf
+++ b/LFM2.5-350M.Q3_K_M.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce630363b2c59015cf5305d0578c05defaa935b8b32a6de8b07da59c39c1dfe4
+size 193151552
--- a/LFM2.5-350M.Q4_K_M.gguf
+++ b/LFM2.5-350M.Q4_K_M.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:021eb5a5241b1ab3dbaec173b465990b0cadd8c65d7babbba41d6ec7d0f6a3e6
+size 229311040
--- a/LFM2.5-350M.Q5_K_M.gguf
+++ b/LFM2.5-350M.Q5_K_M.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f336cebd91677eda4f4ffd30ec1431b6d797900b0d3d5823abc3196a0c414cba
+size 260375104
--- a/LFM2.5-350M.Q6_K.gguf
+++ b/LFM2.5-350M.Q6_K.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bf9fe373661c2642af3bccd96ae3705e7bc26169d7ca9c0acf6a9b8afd70ad3
+size 293380672
--- a/LFM2.5-350M.Q8_0.gguf
+++ b/LFM2.5-350M.Q8_0.gguf
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:476303a70768231f346eadcdf7eec0e984f172addeb3c21833f6422dec500aec
+size 379216448
--- a/README.md
+++ b/README.md
@@ -0,0 +1,199 @@
+---
+language:
+- es
+license: other
+base_model: LiquidAI/LFM2.5-350M
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- base_model:adapter:LiquidAI/LFM2.5-350M
+- lora
+- sft
+- transformers
+- trl
+- unsloth
+---
+
+# simon-fcyt-umss-v2
+
+Modelo de IA ajustado para la Facultad de Ciencias y Tecnología de la Universidad Mayor de San Simón (UMSS), diseñado para la aplicación **TecnoTime**. Su objetivo es ayudar a estudiantes a mantenerse organizados, motivados y conectados con sus actividades académicas mediante respuestas breves, consistentes y parseables.
+
+Este modelo es un fine-tuning de `LiquidAI/LFM2.5-350M` con SFT + LoRA usando Unsloth. Está orientado a producir salidas compatibles con la estructura `SimonResponse` usada por el sistema Android.
+
+Repositorio destino:
+
+```text
+Smith-3/simon-fcyt-umss-v2
+```
+
+## Propósito Académico
+
+El modelo está orientado a:
+
+- Recordatorios de clases y horarios.
+- Mensajes motivacionales para mejorar continuidad académica.
+- Check-ins cortos para acompañamiento del progreso diario.
+- Estímulo positivo en momentos clave del semestre.
+
+Está pensado para reforzar hábitos de constancia, asistencia, estado de ánimo y autocuidado académico.
+
+## Formato Estructurado de Respuesta
+
+La salida esperada no es texto libre. El modelo debe generar un único objeto JSON válido, sin Markdown alrededor, con campos compatibles con TecnoTime.
+
+Campos base:
+
+```json
+{
+  "why": "...",
+  "opening_tag": "...",
+  "icon_tag": "..."
+}
+```
+
+Campos por tipo de notificación:
+
+| Tipo de notificación | Requiere | No debe incluir |
+| --- | --- | --- |
+| `MOTIVATIONAL_CHECK_IN` | `question`, `choices` con 2 opciones | - |
+| `CLASS_REMINDER` | `question`, `choices` con 2 opciones | - |
+| `CHECK_IN_CLOSURE` | `message` o `reminder_copy` | `choices` |
+| `REMINDER_CLOSURE` | `reminder_copy` | `choices` |
+
+## Ejemplos Reales Usados en TecnoTime
+
+### Recordatorio de Clase
+
+```json
+{
+  "why": "Para que no pierdas tu ritmo académico",
+  "opening_tag": "¡Atención, estudiante FCyT!",
+  "question": "¿Listo para tu clase de hoy?",
+  "icon_tag": "info",
+  "choices": [
+    {"id": "yes", "label": "Llego en un momento"},
+    {"id": "no", "label": "Estoy terminando algo"}
+  ]
+}
+```
+
+### Check-in Motivacional
+
+```json
+{
+  "why": "Para reforzar tu constancia",
+  "opening_tag": "¡Vamos con toda, ingeniería!",
+  "question": "¿Cómo te sientes con lo que has avanzado hoy?",
+  "icon_tag": "focus",
+  "choices": [
+    {"id": "yes", "label": "Bien, sigo avanzando"},
+    {"id": "no", "label": "Hoy fue complicado"}
+  ]
+}
+```
+
+## Modelo Base y Entrenamiento
+
+- Modelo base: `LiquidAI/LFM2.5-350M`
+- Tokenizer usado en el notebook: `unsloth/LFM2.5-1.2B-Instruct`
+- Método: SFT + LoRA
+- Framework: Unsloth + TRL + PEFT
+- Dataset: `data/processed/train_data.jsonl`
+- Formato del dataset: LiquidAI `messages`
+
+Configuración de entrenamiento v2:
+
+| Parámetro | Valor |
+| --- | --- |
+| Épocas | 3 |
+| Batch por dispositivo | 4 |
+| Acumulación de gradiente | 4 |
+| Batch efectivo | 16 |
+| Learning rate | `2e-4` |
+| Scheduler | `cosine` |
+| Warmup ratio | `0.03` |
+| LoRA rank | `16` |
+| LoRA alpha | `32` |
+| LoRA dropout | `0.05` |
+
+## Archivos Esperados
+
+Este repositorio está preparado para incluir:
+
+| Archivo o variante | Uso recomendado |
+| --- | --- |
+| Adaptadores LoRA | Respaldo reproducible del fine-tuning |
+| Modelo fusionado 16-bit | Versión principal para despliegue o conversión |
+| `F16` GGUF | Referencia de mayor tamaño |
+| `Q8_0` GGUF | Alta calidad, requiere más RAM |
+| `Q6_K` GGUF | Buen equilibrio con alta calidad |
+| `Q5_K_M` GGUF | Mejor equilibrio calidad/rendimiento |
+| `Q4_K_M` GGUF | Equipos de bajos recursos, laboratorios o laptops modestas |
+| `Q3_K_M` GGUF | Variante más liviana |
+
+## Uso con llama.cpp
+
+Ejemplo con una cuantización GGUF:
+
+```bash
+llama-cli -m simon-fcyt-umss-v2-Q5_K_M.gguf -p "GENERAR: CLASS_REMINDER para clase de Física"
+```
+
+## Uso con Transformers
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+repo_id = "Smith-3/simon-fcyt-umss-v2"
+
+tokenizer = AutoTokenizer.from_pretrained(repo_id)
+model = AutoModelForCausalLM.from_pretrained(repo_id, torch_dtype="auto")
+
+messages = [
+    {
+        "role": "user",
+        "content": "GENERAR: MOTIVATIONAL_CHECK_IN para estudiante de Ingeniería de Sistemas"
+    }
+]
+
+prompt = tokenizer.apply_chat_template(
+    messages,
+    tokenize=False,
+    add_generation_prompt=True,
+)
+
+inputs = tokenizer(prompt, return_tensors="pt")
+outputs = model.generate(**inputs, max_new_tokens=180)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+## Uso Previsto
+
+Uso previsto:
+
+- Integración en TecnoTime.
+- Generación de respuestas JSON para notificaciones académicas.
+- Apoyo motivacional breve para estudiantes.
+- Inferencia local o de bajo costo con GGUF.
+
+Fuera de alcance:
+
+- Información académica oficial sin validación externa.
+- Consejería psicológica, médica o legal.
+- Decisiones administrativas o académicas automatizadas.
+
+## Licencia
+
+Este modelo deriva de `LiquidAI/LFM2.5-350M`. Revisa la licencia del modelo base en Hugging Face antes de redistribuir o usar comercialmente:
+
+https://huggingface.co/LiquidAI/LFM2.5-350M
+
+## Estado
+
+- Adaptadores LoRA v2: preparados para publicación.
+- Modelo fusionado: debe generarse desde el notebook antes de subir.
+- GGUF completo: debe generarse desde el notebook antes de subir.
+- Próxima etapa: publicación e integración final en TecnoTime.
+### Framework versions
+
+- PEFT 0.17.1
--- a/adapter_config.json
+++ b/adapter_config.json
@@ -0,0 +1,47 @@
+{
+  "alpha_pattern": {},
+  "auto_mapping": {
+    "base_model_class": "Lfm2ForCausalLM",
+    "parent_library": "transformers.models.lfm2.modeling_lfm2",
+    "unsloth_fixed": true
+  },
+  "base_model_name_or_path": "LiquidAI/LFM2.5-350M",
+  "bias": "none",
+  "corda_config": null,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 32,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "qalora_group_size": 16,
+  "r": 16,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "out_proj",
+    "v_proj",
+    "w3",
+    "w1",
+    "k_proj",
+    "in_proj",
+    "w2",
+    "q_proj"
+  ],
+  "target_parameters": null,
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}
--- a/adapter_model.safetensors
+++ b/adapter_model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:043af753d7a457c238203ff164306be1b95ccc4ae7398febdbf736a02881502e
+size 24010560
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -0,0 +1,45 @@
+{{- bos_token -}}
+{%- set keep_past_thinking = keep_past_thinking | default(false) -%}
+{%- set ns = namespace(system_prompt="") -%}
+{%- if messages[0]["role"] == "system" -%}
+    {%- set ns.system_prompt = messages[0]["content"] -%}
+    {%- set messages = messages[1:] -%}
+{%- endif -%}
+{%- if tools -%}
+    {%- set ns.system_prompt = ns.system_prompt + ("\n" if ns.system_prompt else "") + "List of tools: [" -%}
+    {%- for tool in tools -%}
+        {%- if tool is not string -%}
+            {%- set tool = tool | tojson -%}
+        {%- endif -%}
+        {%- set ns.system_prompt = ns.system_prompt + tool -%}
+        {%- if not loop.last -%}
+            {%- set ns.system_prompt = ns.system_prompt + ", " -%}
+        {%- endif -%}
+    {%- endfor -%}
+    {%- set ns.system_prompt = ns.system_prompt + "]" -%}
+{%- endif -%}
+{%- if ns.system_prompt -%}
+    {{- "<|im_start|>system\n" + ns.system_prompt + "<|im_end|>\n" -}}
+{%- endif -%}
+{%- set ns.last_assistant_index = -1 -%}
+{%- for message in messages -%}
+    {%- if message["role"] == "assistant" -%}
+        {%- set ns.last_assistant_index = loop.index0 -%}
+    {%- endif -%}
+{%- endfor -%}
+{%- for message in messages -%}
+    {{- "<|im_start|>" + message["role"] + "\n" -}}
+    {%- set content = message["content"] -%}
+    {%- if content is not string -%}
+        {%- set content = content | tojson -%}
+    {%- endif -%}
+    {%- if message["role"] == "assistant" and not keep_past_thinking and loop.index0 != ns.last_assistant_index -%}
+        {%- if "</think>" in content -%}
+            {%- set content = content.split("</think>")[-1] | trim -%}
+        {%- endif -%}
+    {%- endif -%}
+    {{- content + "<|im_end|>\n" -}}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{- "<|im_start|>assistant\n" -}}
+{%- endif -%}
--- a/config.json
+++ b/config.json
@@ -0,0 +1,62 @@
+{
+  "architectures": [
+    "Lfm2ForCausalLM"
+  ],
+  "block_auto_adjust_ff_dim": true,
+  "block_dim": 1024,
+  "block_ff_dim": 6656,
+  "block_ffn_dim_multiplier": 1.0,
+  "block_mlp_init_scale": 1.0,
+  "block_multiple_of": 256,
+  "block_norm_eps": 1e-05,
+  "block_out_init_scale": 1.0,
+  "block_use_swiglu": true,
+  "block_use_xavier_init": true,
+  "bos_token_id": 1,
+  "conv_L_cache": 3,
+  "conv_bias": false,
+  "conv_dim": 1024,
+  "conv_use_xavier_init": true,
+  "torch_dtype": "bfloat16",
+  "eos_token_id": 7,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 6656,
+  "layer_types": [
+    "conv",
+    "conv",
+    "full_attention",
+    "conv",
+    "conv",
+    "full_attention",
+    "conv",
+    "conv",
+    "full_attention",
+    "conv",
+    "full_attention",
+    "conv",
+    "full_attention",
+    "conv",
+    "full_attention",
+    "conv"
+  ],
+  "max_position_embeddings": 128000,
+  "model_type": "lfm2",
+  "norm_eps": 1e-05,
+  "num_attention_heads": 16,
+  "num_heads": 16,
+  "num_hidden_layers": 16,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "rope_parameters": {
+    "rope_theta": 1000000.0,
+    "rope_type": "default"
+  },
+  "rope_theta": 1000000.0,
+  "tie_embedding": true,
+  "transformers_version": "4.57.3",
+  "unsloth_version": "2025.10.9",
+  "use_cache": true,
+  "use_pos_enc": true,
+  "vocab_size": 65536
+}
--- a/model.safetensors
+++ b/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bef2bbe1b8dd3f6b94f038625449d50c2de0a38a0bf72e9c6f0a9b9cede00a8
+size 708984464
--- a/special_tokens_map.json
+++ b/special_tokens_map.json
@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<|startoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|pad|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
--- a/tokenizer.json
+++ b/tokenizer.json
--- a/tokenizer_config.json
+++ b/tokenizer_config.json