Spaces:
Running
on
A10G
Running
on
A10G
Update the default model created with the space. (#74)
Browse files- Update app.py (9b6f1ba8f84e2d08078faacad363a09c8d64f087)
- Update app.py (472bd039f0ae12bb4c305eac4ec023f14757796a)
app.py
CHANGED
@@ -119,31 +119,53 @@ def process_model(model_id, q_method, private_repo, split_model, split_max_tenso
|
|
119 |
card.data.tags = []
|
120 |
card.data.tags.append("llama-cpp")
|
121 |
card.data.tags.append("gguf-my-repo")
|
|
|
|
|
|
|
122 |
card.text = dedent(
|
123 |
f"""
|
124 |
# {new_repo_id}
|
125 |
This model was converted to GGUF format from [`{model_id}`](https://huggingface.co/{model_id}) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
126 |
Refer to the [original model card](https://huggingface.co/{model_id}) for more details on the model.
|
|
|
127 |
## Use with llama.cpp
|
128 |
-
Install llama.cpp through brew
|
|
|
129 |
```bash
|
130 |
-
brew install
|
|
|
131 |
```
|
132 |
Invoke the llama.cpp server or the CLI.
|
133 |
-
|
|
|
134 |
```bash
|
135 |
-
llama
|
136 |
```
|
137 |
-
|
|
|
138 |
```bash
|
139 |
-
llama-server --hf-repo {new_repo_id} --
|
140 |
```
|
|
|
141 |
Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
142 |
```
|
143 |
-
|
144 |
-
cd llama.cpp && \\
|
145 |
-
make && \\
|
146 |
-
./main -m {quantized_gguf_name} -n 128
|
147 |
```
|
148 |
"""
|
149 |
)
|
|
|
119 |
card.data.tags = []
|
120 |
card.data.tags.append("llama-cpp")
|
121 |
card.data.tags.append("gguf-my-repo")
|
122 |
+
if card.data.base_model is None:
|
123 |
+
card.data.base_model = []
|
124 |
+
card.data.base_model.append({model_id})
|
125 |
card.text = dedent(
|
126 |
f"""
|
127 |
# {new_repo_id}
|
128 |
This model was converted to GGUF format from [`{model_id}`](https://huggingface.co/{model_id}) using llama.cpp via the ggml.ai's [GGUF-my-repo](https://huggingface.co/spaces/ggml-org/gguf-my-repo) space.
|
129 |
Refer to the [original model card](https://huggingface.co/{model_id}) for more details on the model.
|
130 |
+
|
131 |
## Use with llama.cpp
|
132 |
+
Install llama.cpp through brew (works on Mac and Linux)
|
133 |
+
|
134 |
```bash
|
135 |
+
brew install llama.cpp
|
136 |
+
|
137 |
```
|
138 |
Invoke the llama.cpp server or the CLI.
|
139 |
+
|
140 |
+
### CLI:
|
141 |
```bash
|
142 |
+
llama --hf-repo {new_repo_id} --hf-file {quantized_gguf_name} -p "The meaning to life and the universe is"
|
143 |
```
|
144 |
+
|
145 |
+
### Server:
|
146 |
```bash
|
147 |
+
llama-server --hf-repo {new_repo_id} --hf-file {quantized_gguf_name} -c 2048
|
148 |
```
|
149 |
+
|
150 |
Note: You can also use this checkpoint directly through the [usage steps](https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#usage) listed in the Llama.cpp repo as well.
|
151 |
+
|
152 |
+
Step 1: Clone llama.cpp from GitHub.
|
153 |
+
```
|
154 |
+
git clone https://github.com/ggerganov/llama.cpp
|
155 |
+
```
|
156 |
+
|
157 |
+
Step 2: Move into the llama.cpp folder and build it with `LLAMA_CURL=1` flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).
|
158 |
+
```
|
159 |
+
cd llama.cpp && LLAMA_CURL=1 make
|
160 |
+
```
|
161 |
+
|
162 |
+
Step 3: Run inference through the main binary.
|
163 |
+
```
|
164 |
+
./main --hf-repo {new_repo_id} --hf-file {quantized_gguf_name} -p "The meaning to life and the universe is"
|
165 |
+
```
|
166 |
+
or
|
167 |
```
|
168 |
+
./server --hf-repo {new_repo_id} --hf-file {quantized_gguf_name} -c 2048
|
|
|
|
|
|
|
169 |
```
|
170 |
"""
|
171 |
)
|