Install from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf fevohh/RayExtract-1B-v0.2:# Run inference directly in the terminal:
llama cli -hf fevohh/RayExtract-1B-v0.2:Use pre-built binary
# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf fevohh/RayExtract-1B-v0.2:# Run inference directly in the terminal:
./llama-cli -hf fevohh/RayExtract-1B-v0.2:Build from source code
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf fevohh/RayExtract-1B-v0.2:# Run inference directly in the terminal:
./build/bin/llama-cli -hf fevohh/RayExtract-1B-v0.2:Use Docker
docker model run hf.co/fevohh/RayExtract-1B-v0.2:YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Remarks:
So far so good, model appears to accurately extract item details for Rayman fist, but it performs poorly on chat and non-rayman advertisment test datasets that it wasnt trained on, but I anticipated this to happen because the v0.2 dataset is an experimental dataset strictly containing only Rayman fist advertisement dataset to test if the finetuned model can perform better than its v0 variant. The next model iteration will be v0.3 with a much higher dataset size, so training duration might be much longer (one pattern that i found is that the v0.2 dataset is about 2x larger than v1 dataset, training time for v0.2 is about 3x longer than training v1/v0 with the same training parameters, i had to manually stop training early when it shows signs of overfitting in v0.2, but regardless it still gives accurate output)
- Downloads last month
- 36
4-bit
5-bit
8-bit
16-bit
Install (macOS, Linux)
# Start a local OpenAI-compatible server with a web UI: llama serve -hf fevohh/RayExtract-1B-v0.2:# Run inference directly in the terminal: llama cli -hf fevohh/RayExtract-1B-v0.2: