blackroad-os/blackroad-ai-integrations
9
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
b619ed617b4b74c439109d97a787e7f87319a196
blackroad-ai-integrations/fireworks-ai.sh

355 lines
11 KiB
Bash
Executable File
Raw Blame History

#!/bin/bash
# Fireworks.ai Integration Script
# Usage: source ~/blackroad-ai-integrations/fireworks-ai.sh
# Provides functions for Fireworks.ai API interactions
set -e
# BlackRoad Brand Colors
PINK='\033[38;5;205m'
AMBER='\033[38;5;214m'
BLUE='\033[38;5;69m'
VIOLET='\033[38;5;135m'
GREEN='\033[38;5;82m'
RED='\033[38;5;196m'
RESET='\033[0m'
# Fireworks API Configuration
FIREWORKS_API_BASE="https://api.fireworks.ai/inference/v1"
FIREWORKS_DEFAULT_MODEL="accounts/fireworks/models/llama-v3p1-70b-instruct"
# Check for API key
fireworks_check_api_key() {
if [ -z "$FIREWORKS_API_KEY" ]; then
echo -e "${RED}ERROR: FIREWORKS_API_KEY environment variable not set${RESET}" >&2
echo -e "${AMBER}Please set it with: export FIREWORKS_API_KEY='your-api-key'${RESET}" >&2
echo -e "${BLUE}Get your API key from: https://fireworks.ai/api-keys${RESET}" >&2
return 1
fi
return 0
}
# Test API connection
fireworks_test_connection() {
echo -e "${BLUE}Testing Fireworks.ai API connection...${RESET}"
if ! fireworks_check_api_key; then
return 1
fi
local response
response=$(curl -s -w "\n%{http_code}" \
-X POST "${FIREWORKS_API_BASE}/chat/completions" \
-H "Authorization: Bearer ${FIREWORKS_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "'"${FIREWORKS_DEFAULT_MODEL}"'",
"messages": [{"role": "user", "content": "Say hello"}],
"max_tokens": 10
}')
local http_code=$(echo "$response" | tail -n1)
local body=$(echo "$response" | sed '$d')
if [ "$http_code" -eq 200 ]; then
echo -e "${GREEN}✓ API connection successful${RESET}"
return 0
else
echo -e "${RED}✗ API connection failed (HTTP $http_code)${RESET}"
echo -e "${AMBER}Response: $body${RESET}"
return 1
fi
}
# List available models
fireworks_list_models() {
if ! fireworks_check_api_key; then
return 1
fi
echo -e "${BLUE}Fetching available Fireworks.ai models...${RESET}"
curl -s -X GET "${FIREWORKS_API_BASE}/models" \
-H "Authorization: Bearer ${FIREWORKS_API_KEY}" \
| jq -r '.data[] | "\(.id) - \(.owned_by)"' 2>/dev/null \
|| echo -e "${RED}Failed to fetch models. Is jq installed?${RESET}"
}
# Text generation (non-streaming)
fireworks_generate() {
local prompt="$1"
local model="${2:-$FIREWORKS_DEFAULT_MODEL}"
local max_tokens="${3:-1024}"
local temperature="${4:-0.7}"
if ! fireworks_check_api_key; then
return 1
fi
if [ -z "$prompt" ]; then
echo -e "${RED}ERROR: Prompt required${RESET}" >&2
echo -e "${AMBER}Usage: fireworks_generate \"your prompt\" [model] [max_tokens] [temperature]${RESET}" >&2
return 1
fi
echo -e "${BLUE}Generating response with Fireworks.ai...${RESET}" >&2
local response
response=$(curl -s -X POST "${FIREWORKS_API_BASE}/chat/completions" \
-H "Authorization: Bearer ${FIREWORKS_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "'"${model}"'",
"messages": [{"role": "user", "content": "'"$(echo "$prompt" | sed 's/"/\\"/g')"'"}],
"max_tokens": '"${max_tokens}"',
"temperature": '"${temperature}"'
}')
# Extract and display the response
local content
content=$(echo "$response" | jq -r '.choices[0].message.content' 2>/dev/null)
if [ "$content" != "null" ] && [ -n "$content" ]; then
echo -e "${GREEN}Response:${RESET}" >&2
echo "$content"
# Display usage stats
local usage
usage=$(echo "$response" | jq -r '.usage' 2>/dev/null)
if [ "$usage" != "null" ]; then
echo -e "\n${VIOLET}Usage: $(echo "$usage" | jq -c .)${RESET}" >&2
fi
else
echo -e "${RED}Failed to get response${RESET}" >&2
echo -e "${AMBER}Raw response: $response${RESET}" >&2
return 1
fi
}
# Streaming text generation
fireworks_generate_stream() {
local prompt="$1"
local model="${2:-$FIREWORKS_DEFAULT_MODEL}"
local max_tokens="${3:-1024}"
local temperature="${4:-0.7}"
if ! fireworks_check_api_key; then
return 1
fi
if [ -z "$prompt" ]; then
echo -e "${RED}ERROR: Prompt required${RESET}" >&2
echo -e "${AMBER}Usage: fireworks_generate_stream \"your prompt\" [model] [max_tokens] [temperature]${RESET}" >&2
return 1
fi
echo -e "${BLUE}Streaming response from Fireworks.ai...${RESET}" >&2
echo -e "${GREEN}Response:${RESET}" >&2
curl -s -N -X POST "${FIREWORKS_API_BASE}/chat/completions" \
-H "Authorization: Bearer ${FIREWORKS_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "'"${model}"'",
"messages": [{"role": "user", "content": "'"$(echo "$prompt" | sed 's/"/\\"/g')"'"}],
"max_tokens": '"${max_tokens}"',
"temperature": '"${temperature}"',
"stream": true
}' | while IFS= read -r line; do
# Parse SSE format (data: {json})
if [[ "$line" == data:* ]]; then
local data="${line#data: }"
if [ "$data" != "[DONE]" ]; then
# Extract delta content
local delta
delta=$(echo "$data" | jq -r '.choices[0].delta.content // empty' 2>/dev/null)
if [ -n "$delta" ] && [ "$delta" != "null" ]; then
printf "%s" "$delta"
fi
fi
fi
done
echo -e "\n${VIOLET}Stream complete${RESET}" >&2
}
# Multi-turn conversation
fireworks_chat() {
local model="${1:-$FIREWORKS_DEFAULT_MODEL}"
if ! fireworks_check_api_key; then
return 1
fi
echo -e "${PINK}=== Fireworks.ai Chat ===${RESET}"
echo -e "${AMBER}Model: $model${RESET}"
echo -e "${BLUE}Type 'exit' or 'quit' to end conversation${RESET}"
echo ""
local messages='[]'
while true; do
echo -ne "${GREEN}You: ${RESET}"
read -r user_input
if [ "$user_input" = "exit" ] || [ "$user_input" = "quit" ]; then
echo -e "${VIOLET}Goodbye!${RESET}"
break
fi
if [ -z "$user_input" ]; then
continue
fi
# Add user message to history
messages=$(echo "$messages" | jq --arg content "$user_input" '. + [{"role": "user", "content": $content}]')
# Get response
echo -ne "${BLUE}Assistant: ${RESET}"
local response
response=$(curl -s -X POST "${FIREWORKS_API_BASE}/chat/completions" \
-H "Authorization: Bearer ${FIREWORKS_API_KEY}" \
-H "Content-Type: application/json" \
-d '{
"model": "'"${model}"'",
"messages": '"${messages}"',
"max_tokens": 1024,
"temperature": 0.7
}')
local assistant_msg
assistant_msg=$(echo "$response" | jq -r '.choices[0].message.content' 2>/dev/null)
if [ "$assistant_msg" != "null" ] && [ -n "$assistant_msg" ]; then
echo "$assistant_msg"
echo ""
# Add assistant message to history
messages=$(echo "$messages" | jq --arg content "$assistant_msg" '. + [{"role": "assistant", "content": $content}]')
else
echo -e "${RED}Error getting response${RESET}"
echo -e "${AMBER}$response${RESET}"
fi
done
}
# Batch processing from file
fireworks_batch() {
local input_file="$1"
local output_file="$2"
local model="${3:-$FIREWORKS_DEFAULT_MODEL}"
if ! fireworks_check_api_key; then
return 1
fi
if [ ! -f "$input_file" ]; then
echo -e "${RED}ERROR: Input file not found: $input_file${RESET}" >&2
return 1
fi
echo -e "${BLUE}Processing batch from $input_file...${RESET}"
> "$output_file" # Clear output file
local count=0
while IFS= read -r prompt; do
if [ -z "$prompt" ]; then
continue
fi
count=$((count + 1))
echo -e "${VIOLET}Processing prompt $count...${RESET}"
echo "=== Prompt $count ===" >> "$output_file"
echo "$prompt" >> "$output_file"
echo "" >> "$output_file"
echo "=== Response ===" >> "$output_file"
fireworks_generate "$prompt" "$model" 512 0.7 2>/dev/null >> "$output_file"
echo -e "\n---\n" >> "$output_file"
done < "$input_file"
echo -e "${GREEN}✓ Processed $count prompts, results saved to $output_file${RESET}"
}
# Function to display help
fireworks_help() {
cat << 'EOF'
Fireworks.ai Integration Functions
====================================
Setup:
export FIREWORKS_API_KEY='your-api-key'
source ~/blackroad-ai-integrations/fireworks-ai.sh
Functions:
fireworks_check_api_key - Verify API key is set
fireworks_test_connection - Test API connectivity
fireworks_list_models - List available models
fireworks_generate <prompt> [model] [max_tokens] [temperature]
- Generate text (non-streaming)
- Example: fireworks_generate "Explain quantum computing"
fireworks_generate_stream <prompt> [model] [max_tokens] [temperature]
- Generate text with streaming
- Example: fireworks_generate_stream "Write a poem about AI"
fireworks_chat [model]
- Interactive chat session
- Example: fireworks_chat
fireworks_batch <input_file> <output_file> [model]
- Process multiple prompts from file
- Example: fireworks_batch prompts.txt results.txt
fireworks_help - Show this help message
Popular Models:
- accounts/fireworks/models/llama-v3p1-70b-instruct (default)
- accounts/fireworks/models/llama-v3p1-8b-instruct
- accounts/fireworks/models/mixtral-8x7b-instruct
- accounts/fireworks/models/qwen2p5-72b-instruct
Environment Variables:
FIREWORKS_API_KEY - Your Fireworks.ai API key (required)
FIREWORKS_DEFAULT_MODEL - Override default model
Examples:
# Basic generation
fireworks_generate "What is the capital of France?"
# Streaming response
fireworks_generate_stream "Write a short story"
# Use specific model
fireworks_generate "Explain AI" "accounts/fireworks/models/llama-v3p1-8b-instruct"
# Interactive chat
fireworks_chat
# Batch processing
echo -e "What is 2+2?\nWhat is the speed of light?" > prompts.txt
fireworks_batch prompts.txt results.txt
Get API Key:
https://fireworks.ai/api-keys
EOF
}
# Auto-display help if sourced directly
if [ "${BASH_SOURCE[0]}" = "${0}" ]; then
echo -e "${PINK}Fireworks.ai Integration${RESET}"
echo -e "${AMBER}This script should be sourced, not executed directly${RESET}"
echo -e "${BLUE}Usage: source ~/blackroad-ai-integrations/fireworks-ai.sh${RESET}"
echo ""
fireworks_help
else
echo -e "${GREEN}✓ Fireworks.ai integration loaded${RESET}"
echo -e "${BLUE}Run 'fireworks_help' for usage information${RESET}"
fi
Reference in New Issue View Git Blame Copy Permalink