初始化项目，由ModelHub XC社区提供模型

Model: NeshVerse/Flash-financial-analysis-lfm-1.2b
Source: Original Platform

.gitattributes

@@ -0,0 +1,2 @@
+model.safetensors filter=lfs diff=lfs merge=lfs -text
+flash-financial-analysis-q8_0-v2.gguf filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,109 @@
+---
+license: apache-2.0
+language:
+  - en
+base_model: LiquidAI/LFM2.5-1.2B-Base
+tags:
+  - finance
+  - sales-analysis
+  - structured-data
+  - unsloth
+  - lora
+  - lfm
+  - data-analysis
+  - time-series
+  - business-intelligence
+  - fast-inference
+  - efficient-finetuning
+pipeline_tag: text-generation
+library_name: transformers
+---
+
+# ⚡ Flash-Financial-Analysis-LFM-1.2B
+
+**Lightning-fast financial intelligence for structured data analysis**
+
+A blazing-fast, customized, lightweight language model optimized for real-time sales & stock analytics, inventory insights, and financial reporting based on the LiquidAI 1.2B base model supervised fine-tuned FP16 model.
+
+## Rag Model Github Repo:
+https://github.com/neshverse/Flash-RAG-web-GUI/tree/main
+
+## Space (Real-time Testing) 
+https://huggingface.co/spaces/NeshVerse/Flash-financial-analysis-lfm-1.2b
+
+Note: If you get any error while chatting, just refresh the page.
+
+## Model Details
+
+| Attribute | Value |
+|-----------|-------|
+| **Base Architecture** | [LiquidAI/LFM2.5-1.2B-Base](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Base) |
+| **Fine-tuning** | LoRA (r=4, alpha=8) |
+| **Context Window** | 1,024 tokens |
+| **Precision** | FP16 |
+| **Parameters** | 1.2B base + ~500K LoRA |
+
+## Training Summary
+
+- **Total Samples**: 39,435 (37,463 train / 1,972 validation)
+- **Training Duration**: 2.4 hours
+- **Final Loss**: 0.497 (train) / 0.508 (validation)
+- **Hardware**: Consumer GPU (T4)
+
+## Capabilities
+
+- **Sales Analytics**: Real-time sales data querying and analysis
+- **Stock Analytics**: Inventory levels, turnover rates, stock movement
+- **Financial Reporting**: Automated report generation from structured data
+- **Inventory Insights**: Product performance, seasonal trends, demand forecasting
+
+## Performance
+
+- **Inference Speed**: ~0.55 it/s (T4 GPU)
+- **Memory Usage**: ~6GB (4-bit loaded)
+- **Batch Size**: 4 (effective 8 with grad accum)
+- **Max Sequence**: 1,024 tokens
+  
+## Limitations
+
+- Optimized for structured financial/sales data queries
+- Context window limited to 1,024 tokens
+- Training data from 2022-2023; may not reflect current market conditions
+- Best performance on English language inputs
+
+## Model Files
+
+| File | Format | Size | Description | Use Case |
+|------|--------|------|-------------|----------|
+| `model.safetensors` | FP16 | ~2.4 GB | Original full precision | Maximum quality, GPU inference |
+| `flash-financial-analysis-q8_0.gguf` | Q8_0 | ~1.2 GB | 8-bit quantized (llama.cpp) | CPU inference, Ollama, LM Studio |
+
+## Quantized Version (Q8_0)
+
+We now provide a **Q8_0 quantized version** for easier deployment:
+
+- **Format**: GGUF (llama.cpp compatible)
+- **Size**: ~50% smaller than FP16 (1.2 GB vs 2.4 GB)
+- **Quality**: ~99.9% of original performance
+- **Tools**: Works with llama.cpp, Ollama, LM Studio, llama-cpp-python
+
+### Download Q8_0
+
+```bash
+# Using huggingface-cli
+huggingface-cli download NeshVerse/Flash-financial-analysis-lfm-1.2b flash-financial-analysis-q8_0.gguf
+
+# Or direct download
+wget https://huggingface.co/NeshVerse/Flash-financial-analysis-lfm-1.2b/resolve/main/flash-financial-analysis-q8_0.gguf
+
+## Quick Start
+
+```python
+from unsloth import FastLanguageModel
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+    "NeshVerse/Flash-financial-analysis-lfm-1.2b",
+    max_seq_length=1024,
+    load_in_4bit=True,
+    trust_remote_code=True,
+)

chat_template.jinja

@@ -0,0 +1,7 @@
+{{- bos_token -}}{%- set system_prompt = "" -%}{%- 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 + ("
+" if ns.system_prompt else "") + "List of tools: <|tool_list_start|>[" -%} {%- 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 + "]<|tool_list_end|>" -%}{%- endif -%}{%- if ns.system_prompt -%} {{- "<|im_start|>system
+" + ns.system_prompt + "<|im_end|>
+" -}}{%- endif -%}{%- for message in messages -%} {{- "<|im_start|>" + message["role"] + "
+" -}} {%- set content = message["content"] -%} {%- if content is not string -%} {%- set content = content | tojson -%} {%- endif -%} {%- if message["role"] == "tool" -%} {%- set content = "<|tool_response_start|>" + content + "<|tool_response_end|>" -%} {%- endif -%} {{- content + "<|im_end|>
+" -}}{%- endfor -%}{%- if add_generation_prompt -%} {{- "<|im_start|>assistant
+" -}}{%- endif -%}

config.json

@@ -0,0 +1,62 @@
+{
+  "architectures": [
+    "Lfm2ForCausalLM"
+  ],
+  "block_auto_adjust_ff_dim": true,
+  "block_dim": 2048,
+  "block_ff_dim": 12288,
+  "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": 2048,
+  "conv_use_xavier_init": true,
+  "dtype": "float16",
+  "eos_token_id": 7,
+  "hidden_size": 2048,
+  "initializer_range": 0.02,
+  "intermediate_size": 12288,
+  "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": 32,
+  "num_heads": 32,
+  "num_hidden_layers": 16,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "rope_parameters": {
+    "rope_theta": 1000000.0,
+    "rope_type": "default"
+  },
+  "tie_embedding": true,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.1.0",
+  "unsloth_version": "2026.2.1",
+  "use_cache": true,
+  "use_pos_enc": true,
+  "vocab_size": 65536
+}

flash-financial-analysis-q8_0-v2.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d58d2595df10e711baeb916ce39bedb90d894aa6f5e3592f376699f8936d924d
+size 1246252928

generation_config.json

@@ -0,0 +1,8 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 7,
+  "max_length": 128000,
+  "pad_token_id": 0,
+  "transformers_version": "5.1.0"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36bc54c83905c9157aa3739ba6966954d3487f6e554ebfc993d0674827638096
+size 2340697784

tokenizer_config.json

@@ -0,0 +1,20 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<|startoftext|>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "is_local": false,
+  "legacy": false,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<|pad|>",
+  "padding_side": "left",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "TokenizersBackend",
+  "use_default_system_prompt": false,
+  "use_fast": true
+}