Nafiul Huda commited on
Update Dockerfile
Browse files- Dockerfile +6 -7
Dockerfile
CHANGED
|
@@ -1,17 +1,16 @@
|
|
| 1 |
FROM ghcr.io/ggml-org/llama.cpp:full
|
| 2 |
-
|
| 3 |
WORKDIR /app
|
| 4 |
|
| 5 |
# প্রয়োজনীয় প্যাকেজ ইনস্টল
|
| 6 |
-
RUN apt update && apt install -y python3-pip
|
| 7 |
-
|
| 8 |
|
| 9 |
-
# Hugging Face থেকে মডেল ডাউনলোড
|
| 10 |
RUN python3 -c 'from huggingface_hub import hf_hub_download; \
|
| 11 |
repo="unsloth/LFM2.5-1.2B-Instruct-GGUF"; \
|
| 12 |
-
hf_hub_download(repo_id=repo, filename="LFM2.5-1.2B-Instruct-UD-Q8_K_XL.gguf", local_dir="/app");'
|
| 13 |
|
| 14 |
-
# সার্ভার কনফিগারেশন
|
| 15 |
ENTRYPOINT ["/app/llama-server"]
|
| 16 |
CMD ["-m", "/app/LFM2.5-1.2B-Instruct-UD-Q8_K_XL.gguf", \
|
| 17 |
"--host", "0.0.0.0", \
|
|
@@ -20,4 +19,4 @@ CMD ["-m", "/app/LFM2.5-1.2B-Instruct-UD-Q8_K_XL.gguf", \
|
|
| 20 |
"--cache-type-k", "q4_0", \
|
| 21 |
"--cache-type-v", "iq4_nl", \
|
| 22 |
"-c", "128000", \
|
| 23 |
-
"-n", "38912"]
|
|
|
|
| 1 |
FROM ghcr.io/ggml-org/llama.cpp:full
|
|
|
|
| 2 |
WORKDIR /app
|
| 3 |
|
| 4 |
# প্রয়োজনীয় প্যাকেজ ইনস্টল
|
| 5 |
+
RUN apt update && apt install -y python3-pip && \
|
| 6 |
+
pip install -U huggingface_hub
|
| 7 |
|
| 8 |
+
# Hugging Face থেকে মডেল ডাউনলোড (সিমলিঙ্ক ছাড়া সরাসরি ডাউনলোড)
|
| 9 |
RUN python3 -c 'from huggingface_hub import hf_hub_download; \
|
| 10 |
repo="unsloth/LFM2.5-1.2B-Instruct-GGUF"; \
|
| 11 |
+
hf_hub_download(repo_id=repo, filename="LFM2.5-1.2B-Instruct-UD-Q8_K_XL.gguf", local_dir="/app", local_dir_use_symlinks=False);'
|
| 12 |
|
| 13 |
+
# সার্ভার কনফিগারেশন
|
| 14 |
ENTRYPOINT ["/app/llama-server"]
|
| 15 |
CMD ["-m", "/app/LFM2.5-1.2B-Instruct-UD-Q8_K_XL.gguf", \
|
| 16 |
"--host", "0.0.0.0", \
|
|
|
|
| 19 |
"--cache-type-k", "q4_0", \
|
| 20 |
"--cache-type-v", "iq4_nl", \
|
| 21 |
"-c", "128000", \
|
| 22 |
+
"-n", "38912"]
|