Safetensors
GGUF
Turkish
llama
Llama-3
instruct
finetune
chatml
gpt4
synthetic data
distillation
function calling
json mode
axolotl
roleplaying
chat
Instructions to use tda45/TdAI with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use tda45/TdAI with llama.cpp:
Install (macOS, Linux)
curl -LsSf https://llama.app/install.sh | sh # Start a local OpenAI-compatible server with a web UI: llama serve -hf tda45/TdAI # Run inference directly in the terminal: llama cli -hf tda45/TdAI
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf tda45/TdAI # Run inference directly in the terminal: llama cli -hf tda45/TdAI
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 tda45/TdAI # Run inference directly in the terminal: ./llama-cli -hf tda45/TdAI
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 tda45/TdAI # Run inference directly in the terminal: ./build/bin/llama-cli -hf tda45/TdAI
Use Docker
docker model run hf.co/tda45/TdAI
- LM Studio
- Jan
- Ollama
How to use tda45/TdAI with Ollama:
ollama run hf.co/tda45/TdAI
- Unsloth Desktop
- Docker Model Runner
How to use tda45/TdAI with Docker Model Runner:
docker model run hf.co/tda45/TdAI
- Lemonade
How to use tda45/TdAI with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull tda45/TdAI
Run and chat with the model
lemonade run user.TdAI-{{QUANT_TAG}}List all available models
lemonade list
- Atomic Chat
File size: 1,475 Bytes
c9c3262 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 | #version 450
#include "dequant_head.glsl"
layout(local_size_x = 64, local_size_y = 1, local_size_z = 1) in;
layout (binding = 0) readonly buffer A {A_TYPE data_a[];};
layout (binding = 1) writeonly buffer D {D_TYPE data_b[];};
void main() {
[[unroll]] for (uint wgy = 0; wgy < 256; wgy++) {
const uint i = gl_WorkGroupID.x * 256 + wgy;
if (i >= p.nel / QUANT_K) {
return;
}
const uint tid = gl_LocalInvocationID.x;
const uint ip = tid / 32;
const uint il = tid - 32 * ip;
const uint is = 8 * ip + il / 16;
const uint y_idx = i * QUANT_K + 128 * ip + il;
const uint ql_idx = 32 * ip + il;
const uint8_t qs = data_a[i].qs[32 * ip + il];
FLOAT_TYPE dall = FLOAT_TYPE(data_a[i].dm.x);
FLOAT_TYPE dmin = FLOAT_TYPE(data_a[i].dm.y);
data_b[y_idx + 0] = D_TYPE(dall * FLOAT_TYPE((data_a[i].scales[is+0] & 0xF) * ((qs >> 0) & 3)) - dmin * FLOAT_TYPE(data_a[i].scales[is+0] >> 4));
data_b[y_idx + 32] = D_TYPE(dall * FLOAT_TYPE((data_a[i].scales[is+2] & 0xF) * ((qs >> 2) & 3)) - dmin * FLOAT_TYPE(data_a[i].scales[is+2] >> 4));
data_b[y_idx + 64] = D_TYPE(dall * FLOAT_TYPE((data_a[i].scales[is+4] & 0xF) * ((qs >> 4) & 3)) - dmin * FLOAT_TYPE(data_a[i].scales[is+4] >> 4));
data_b[y_idx + 96] = D_TYPE(dall * FLOAT_TYPE((data_a[i].scales[is+6] & 0xF) * ((qs >> 6) & 3)) - dmin * FLOAT_TYPE(data_a[i].scales[is+6] >> 4));
}
}
|