DEFAULT_NUM_CTX additions

adding further changes for DEFAULT_NUM_CTX, including docs
This commit is contained in:
Aaron Bolton 2024-11-19 07:46:51 +00:00
parent bfaaf86c69
commit 5359423481
4 changed files with 26 additions and 3 deletions

View File

@ -62,4 +62,5 @@ VITE_LOG_LEVEL=debug
# DEFAULT_NUM_CTX=32768 # Consumes 36GB of VRAM
# DEFAULT_NUM_CTX=24576 # Consumes 32GB of VRAM
# DEFAULT_NUM_CTX=12288 # Consumes 26GB of VRAM
# DEFAULT_NUM_CTX=6144 # Consumes 24GB of VRAM
# DEFAULT_NUM_CTX=6144 # Consumes 24GB of VRAM
DEFAULT_NUM_CTX=

View File

@ -1,4 +1,7 @@
# Contributing to Bolt.new Fork
## DEFAULT_NUM_CTX
The `DEFAULT_NUM_CTX` environment variable can be used to limit the maximum number of context values used by the qwen2.5-coder model. For example, to limit the context to 24576 values (which uses 32GB of VRAM), set `DEFAULT_NUM_CTX=24576` in your `.env.local` file.
First off, thank you for considering contributing to Bolt.new! This fork aims to expand the capabilities of the original project by integrating multiple LLM providers and enhancing functionality. Every contribution helps make Bolt.new a better tool for developers worldwide.
@ -80,6 +83,19 @@ ANTHROPIC_API_KEY=XXX
```bash
VITE_LOG_LEVEL=debug
```
- Optionally set context size:
```bash
DEFAULT_NUM_CTX=32768
```
Some Example Context Values for the qwen2.5-coder:32b models are.
* DEFAULT_NUM_CTX=32768 - Consumes 36GB of VRAM
* DEFAULT_NUM_CTX=24576 - Consumes 32GB of VRAM
* DEFAULT_NUM_CTX=12288 - Consumes 26GB of VRAM
* DEFAULT_NUM_CTX=6144 - Consumes 24GB of VRAM
**Important**: Never commit your `.env.local` file to version control. It's already included in .gitignore.
### 🚀 Running the Development Server

View File

@ -25,6 +25,7 @@ ARG OPEN_ROUTER_API_KEY
ARG GOOGLE_GENERATIVE_AI_API_KEY
ARG OLLAMA_API_BASE_URL
ARG VITE_LOG_LEVEL=debug
ARG DEFAULT_NUM_CTX
ENV WRANGLER_SEND_METRICS=false \
GROQ_API_KEY=${GROQ_API_KEY} \
@ -33,7 +34,8 @@ ENV WRANGLER_SEND_METRICS=false \
OPEN_ROUTER_API_KEY=${OPEN_ROUTER_API_KEY} \
GOOGLE_GENERATIVE_AI_API_KEY=${GOOGLE_GENERATIVE_AI_API_KEY} \
OLLAMA_API_BASE_URL=${OLLAMA_API_BASE_URL} \
VITE_LOG_LEVEL=${VITE_LOG_LEVEL}
VITE_LOG_LEVEL=${VITE_LOG_LEVEL} \
DEFAULT_NUM_CTX=${DEFAULT_NUM_CTX}
# Pre-configure wrangler to disable metrics
RUN mkdir -p /root/.config/.wrangler && \
@ -54,6 +56,7 @@ ARG OPEN_ROUTER_API_KEY
ARG GOOGLE_GENERATIVE_AI_API_KEY
ARG OLLAMA_API_BASE_URL
ARG VITE_LOG_LEVEL=debug
ARG DEFAULT_NUM_CTX
ENV GROQ_API_KEY=${GROQ_API_KEY} \
OPENAI_API_KEY=${OPENAI_API_KEY} \
@ -61,7 +64,8 @@ ENV GROQ_API_KEY=${GROQ_API_KEY} \
OPEN_ROUTER_API_KEY=${OPEN_ROUTER_API_KEY} \
GOOGLE_GENERATIVE_AI_API_KEY=${GOOGLE_GENERATIVE_AI_API_KEY} \
OLLAMA_API_BASE_URL=${OLLAMA_API_BASE_URL} \
VITE_LOG_LEVEL=${VITE_LOG_LEVEL}
VITE_LOG_LEVEL=${VITE_LOG_LEVEL} \
DEFAULT_NUM_CTX=${DEFAULT_NUM_CTX}
RUN mkdir -p ${WORKDIR}/run
CMD pnpm run dev --host

View File

@ -20,6 +20,7 @@ services:
- GOOGLE_GENERATIVE_AI_API_KEY=${GOOGLE_GENERATIVE_AI_API_KEY}
- OLLAMA_API_BASE_URL=${OLLAMA_API_BASE_URL}
- VITE_LOG_LEVEL=${VITE_LOG_LEVEL:-debug}
- DEFAULT_NUM_CTX=${DEFAULT_NUM_CTX:-32768}
- RUNNING_IN_DOCKER=true
extra_hosts:
- "host.docker.internal:host-gateway"
@ -46,6 +47,7 @@ services:
- GOOGLE_GENERATIVE_AI_API_KEY=${GOOGLE_GENERATIVE_AI_API_KEY}
- OLLAMA_API_BASE_URL=${OLLAMA_API_BASE_URL}
- VITE_LOG_LEVEL=${VITE_LOG_LEVEL:-debug}
- DEFAULT_NUM_CTX=${DEFAULT_NUM_CTX:-32768}
- RUNNING_IN_DOCKER=true
extra_hosts:
- "host.docker.internal:host-gateway"