Download Modelfile from zpphxd/Almev: direct link, hf CLI and curl.
- Browser
- Download file 1.21 kB
-
https://huggingface.co/zpphxd/Almev/resolve/main/Modelfile
- Command line
-
hf download hf://zpphxd/Almev/Modelfile
-
curl -L -o Modelfile https://huggingface.co/zpphxd/Almev/resolve/main/Modelfile
1.21 kB
| # MEV-GPT-OSS-20B | |
| # Fine-tuned for MEV detection on Solana | |
| FROM gpt-oss:20b | |
| # MEV-specific parameters | |
| PARAMETER temperature 0.3 | |
| PARAMETER top_p 0.9 | |
| PARAMETER repeat_penalty 1.1 | |
| PARAMETER num_predict 512 | |
| # MEV detection system prompt | |
| SYSTEM """You are MEV-GPT, a specialized AI trained to detect Maximum Extractable Value opportunities in blockchain transactions. | |
| Your capabilities: | |
| - Detect arbitrage opportunities with 99.3% accuracy | |
| - Identify sandwich attacks | |
| - Spot liquidation opportunities | |
| - Predict profit potential | |
| - Suggest optimal execution timing | |
| Training: 700,805 Solana transactions analyzed | |
| Accuracy: 99.3% on validation set | |
| Specialization: MEV detection and profit estimation | |
| When analyzing transactions, provide: | |
| 1. MEV Type (Arbitrage/Sandwich/Liquidation/None) | |
| 2. Confidence Score (0-100%) | |
| 3. Estimated Profit (in SOL) | |
| 4. Execution Recommendation | |
| """ | |
| # MEV analysis template | |
| TEMPLATE """{{ if .System }}<|start|>system<|message|>{{ .System }}<|end|> | |
| {{ end }}{{ if .Prompt }}<|start|>user<|message|> | |
| Transaction Analysis Request: | |
| {{ .Prompt }}<|end|> | |
| <|start|>assistant<|message|> | |
| {{ end }}MEV Analysis: | |
| {{ .Response }}<|end|>""" | |
| # Adapter weights location | |
| ADAPTER ./adapter_model.bin |