rl-agent/.env.example at main · advikdivekar/rl-agent · GitHub

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
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
# ── OPTION 1: HuggingFace Router (default) ───────────────────
# Get token at: https://huggingface.co/settings/tokens
# API_BASE_URL=https://router.huggingface.co/v1
# HF_TOKEN=hf_your_token_here
# MODEL_NAME=Qwen/Qwen2.5-7B-Instruct


# ── OPTION 2: OpenRouter ──────────────────────────────────────
# Get key at: https://openrouter.ai/settings/keys
# API_BASE_URL=https://openrouter.ai/api/v1
# HF_TOKEN=sk-or-your_key_here
# MODEL_NAME=mistralai/mistral-7b-instruct


# ── OPTION 3: NVIDIA NIM ──────────────────────────────────────
# Get key at: https://build.nvidia.com (click any model → Get API Key)
# API_BASE_URL=https://integrate.api.nvidia.com/v1
# HF_TOKEN=nvapi-your_key_here
# MODEL_NAME=meta/llama-3.1-8b-instruct


# ── OPTION 4: OpenAI ─────────────────────────────────────────
# Get key at: https://platform.openai.com/api-keys
# API_BASE_URL=https://api.openai.com/v1
# HF_TOKEN=sk-your_openai_key_here
# MODEL_NAME=gpt-4o-mini


# ── OPTION 5: Groq ───────────────────────────────────────────
# Get key at: https://console.groq.com/keys
# API_BASE_URL=https://api.groq.com/openai/v1
# HF_TOKEN=gsk_your_groq_key_here
# MODEL_NAME=llama-3.3-70b-versatile


# ── OPTION 6: Together AI ────────────────────────────────────
# Get key at: https://api.together.xyz/settings/api-keys
# API_BASE_URL=https://api.together.xyz/v1
# HF_TOKEN=your_together_key_here
# MODEL_NAME=meta-llama/Llama-3-70b-chat-hf


# ── ACTIVE CONFIGURATION (uncomment one block above, copy here) ──
API_BASE_URL=https://router.huggingface.co/v1
HF_TOKEN=your_token_here
MODEL_NAME=Qwen/Qwen2.5-7B-Instruct


# ── ENVIRONMENT SERVER ────────────────────────────────────────
# Local:    http://localhost:7860
# HF Space: https://advikdivekar-scheme-enrollment-env.hf.space
ENV_URL=http://localhost:7860


# ── INFERENCE SETTINGS ────────────────────────────────────────
# Episodes per task — 3 for stable scores, 1 for quick test
N_REPEATS=3

# Sampling temperature — 0.0 for deterministic results
INFERENCE_TEMPERATURE=0.0

# Max tokens per model call
MAX_TOKENS=1500


# ── OPTIONAL ──────────────────────────────────────────────────
# Only needed for from_docker_image() workflows
# LOCAL_IMAGE_NAME=