Instructions to use FINAL-Bench/Aether-7B-5Attn with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use FINAL-Bench/Aether-7B-5Attn with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="FINAL-Bench/Aether-7B-5Attn", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("FINAL-Bench/Aether-7B-5Attn", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use FINAL-Bench/Aether-7B-5Attn with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "FINAL-Bench/Aether-7B-5Attn"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/Aether-7B-5Attn",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/FINAL-Bench/Aether-7B-5Attn

SGLang

How to use FINAL-Bench/Aether-7B-5Attn with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "FINAL-Bench/Aether-7B-5Attn" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/Aether-7B-5Attn",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "FINAL-Bench/Aether-7B-5Attn" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "FINAL-Bench/Aether-7B-5Attn",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use FINAL-Bench/Aether-7B-5Attn with Docker Model Runner:
```
docker model run hf.co/FINAL-Bench/Aether-7B-5Attn
```

SeaWolf-AI commited on 10 days ago

Commit

2cbba77

verified ·

1 Parent(s): f7e2867

sanitize: internal paths/ids/comments

Browse files

Files changed (1) hide show

training/launcher_v2b_multi.py +5 -5

training/launcher_v2b_multi.py CHANGED Viewed

@@ -6,9 +6,9 @@ AETHER-2B-7Attn Phase 1.6 v2b — Continued Pretraining
 Resume:  step_0455000 (6.59B model, 59.8B tokens done)
 Data:    132B pool (phase15_mix 90B + tokenized_v2 42.5B, weighted)
 Goal:    annealing on high-quality data → 132B chinchilla total
-Infra:   multi node 2x8 B200 (CID=5352+5353, 16 GPU)
 LR:      cosine annealing 5e-5 → 5e-6 over 324K steps (~85B new tokens)
-Log:     /NHNHOME/WORKSPACE/0426030036_A/aether_1b_7attn_v3/logs/launcher_v2b.log
 """
 import math, os, sys, time, signal, json, logging
@@ -29,13 +29,13 @@ from torch.distributed.fsdp import (
 from torch.distributed.fsdp.wrap import transformer_auto_wrap_policy
 # ── paths ────────────────────────────────────────────────────────────────────
-BASE      = "/NHNHOME/WORKSPACE/0426030036_A/aether_1b_7attn_v3"
-V2_ROOT   = "/NHNHOME/WORKSPACE/0426030036_A/aether-recovered/aether-v2-current"
 RESUME_CKPT = f"{BASE}/runs/v2b_phase16/ckpt/step_0001905"
 OUTPUT_DIR  = f"{BASE}/runs/v2b_multi_phase16"
 CKPT_DIR    = f"{OUTPUT_DIR}/ckpt"
 LOG_FILE    = f"{BASE}/logs/launcher_v2b_multi.log"
-KILL_SWITCH = "/NHNHOME/STOP_TRAINING_V2B_MULTI"
 sys.path.insert(0, V2_ROOT)
 sys.path.insert(0, V2_ROOT + "/scripts")

 Resume:  step_0455000 (6.59B model, 59.8B tokens done)
 Data:    132B pool (phase15_mix 90B + tokenized_v2 42.5B, weighted)
 Goal:    annealing on high-quality data → 132B chinchilla total
+Infra:   multi node 2x8 B200 (16 GPU, 2-node)
 LR:      cosine annealing 5e-5 → 5e-6 over 324K steps (~85B new tokens)
+Log:     /workspace/aether/logs/launcher_v2b.log
 """
 import math, os, sys, time, signal, json, logging
 from torch.distributed.fsdp.wrap import transformer_auto_wrap_policy
 # ── paths ────────────────────────────────────────────────────────────────────
+BASE      = "/workspace/aether"
+V2_ROOT   = "/workspace/aether-src"
 RESUME_CKPT = f"{BASE}/runs/v2b_phase16/ckpt/step_0001905"
 OUTPUT_DIR  = f"{BASE}/runs/v2b_multi_phase16"
 CKPT_DIR    = f"{OUTPUT_DIR}/ckpt"
 LOG_FILE    = f"{BASE}/logs/launcher_v2b_multi.log"
+KILL_SWITCH = "/workspace/STOP_TRAINING_V2B_MULTI"
 sys.path.insert(0, V2_ROOT)
 sys.path.insert(0, V2_ROOT + "/scripts")