Instructions to use keepsloading/icml_repro_scratch with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use keepsloading/icml_repro_scratch with Transformers:

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

pipe = pipeline("text-generation", model="keepsloading/icml_repro_scratch")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("keepsloading/icml_repro_scratch")
model = AutoModelForCausalLM.from_pretrained("keepsloading/icml_repro_scratch", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use keepsloading/icml_repro_scratch with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/keepsloading/icml_repro_scratch

SGLang

How to use keepsloading/icml_repro_scratch 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 "keepsloading/icml_repro_scratch" \
    --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": "keepsloading/icml_repro_scratch",
		"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 "keepsloading/icml_repro_scratch" \
        --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": "keepsloading/icml_repro_scratch",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use keepsloading/icml_repro_scratch with Docker Model Runner:
```
docker model run hf.co/keepsloading/icml_repro_scratch
```

keepsloading commited on 26 days ago

Commit

5c2f3ce

verified ·

1 Parent(s): 5505109

Upload folder using huggingface_hub

Browse files

Files changed (5) hide show

recipe/data/eval_inputs/helmet_icl_8k_n50_per_config.jsonl +0 -0
recipe/data/eval_inputs/mrcr_8k_2_4_8needle_n10_per_config.jsonl +0 -0
recipe/scripts/eval_sweep.sh +1 -2
recipe/scripts/summarize_qwen1p7b_router_granularity_20260714.py +1 -1
recipe/scripts/train_l2a_only.sh +17 -0

recipe/data/eval_inputs/helmet_icl_8k_n50_per_config.jsonl CHANGED Viewed

The diff for this file is too large to render. See raw diff

recipe/data/eval_inputs/mrcr_8k_2_4_8needle_n10_per_config.jsonl CHANGED Viewed

The diff for this file is too large to render. See raw diff

recipe/scripts/eval_sweep.sh CHANGED Viewed

@@ -16,7 +16,7 @@ IFS=',' read -r -a GPUS <<< "$GPU_LIST"
 declare -A SLOT_PIDS=()
 mkdir -p "$EVAL_ROOT"
-for model in "$VANILLA_MODEL" "$AHA_MODEL" "$L2A_MODEL"; do
   [[ -f "$model/config.json" ]] || { echo "Missing model: $model" >&2; exit 2; }
 done
@@ -27,7 +27,6 @@ done
 for threshold in "${THRESHOLDS[@]}"; do
   slug="${threshold/./}"
   for benchmark in "${BENCHMARKS[@]}"; do
-    jobs+=("token_kv_head_t${slug}|$AHA_MODEL|$threshold|$benchmark")
     jobs+=("token_t${slug}|$L2A_MODEL|$threshold|$benchmark")
   done
 done

 declare -A SLOT_PIDS=()
 mkdir -p "$EVAL_ROOT"
+for model in "$VANILLA_MODEL" "$L2A_MODEL"; do
   [[ -f "$model/config.json" ]] || { echo "Missing model: $model" >&2; exit 2; }
 done
 for threshold in "${THRESHOLDS[@]}"; do
   slug="${threshold/./}"
   for benchmark in "${BENCHMARKS[@]}"; do
     jobs+=("token_t${slug}|$L2A_MODEL|$threshold|$benchmark")
   done
 done

recipe/scripts/summarize_qwen1p7b_router_granularity_20260714.py CHANGED Viewed

@@ -17,7 +17,7 @@ import matplotlib.pyplot as plt
 SUITES = ("RULER-local-full13", "BabiLong", "HELMET-ICL", "MRCR")
 THRESHOLDS = (0.45, 0.50, 0.525, 0.55, 0.575, 0.60, 0.625, 0.65)
-ARMS = ("token", "token_kv_head")
 SLUGS = {
     "RULER-local-full13": "ruler",
     "BabiLong": "babilong",

 SUITES = ("RULER-local-full13", "BabiLong", "HELMET-ICL", "MRCR")
 THRESHOLDS = (0.45, 0.50, 0.525, 0.55, 0.575, 0.60, 0.625, 0.65)
+ARMS = ("token",)
 SLUGS = {
     "RULER-local-full13": "ruler",
     "BabiLong": "babilong",

recipe/scripts/train_l2a_only.sh ADDED Viewed

	@@ -0,0 +1,17 @@

+#!/usr/bin/env bash
+set -euo pipefail
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+RECIPE="$(cd "$SCRIPT_DIR/.." && pwd)"
+GPU_L2A="${GPU_L2A:-0}"
+cd "$RECIPE"
+echo "L2A-style = per-token, head-shared gate; assigned visible GPU $GPU_L2A."
+echo "Training L2A-style variant only."
+python "$SCRIPT_DIR/validate_release.py"
+python -m unittest -v tests.test_router_granularity
+"$SCRIPT_DIR/train_one.sh" l2a_style "$GPU_L2A"
+echo "L2A-style selected checkpoint: ${OUTPUT_ROOT:-$RECIPE/../outputs}/l2a_style/stage2/checkpoint-25"