Text Generation
Transformers
Safetensors
English
Korean
code
fuse_glm
custom_code
lfm2
glm
mixture-of-experts
routed-experts
coding
code-generation
agentic
bf16
top-k-routing
trust-remote-code
conversational
Instructions to use HCHs/RivetCoder-9B-A4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use HCHs/RivetCoder-9B-A4B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="HCHs/RivetCoder-9B-A4B", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("HCHs/RivetCoder-9B-A4B", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use HCHs/RivetCoder-9B-A4B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "HCHs/RivetCoder-9B-A4B" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "HCHs/RivetCoder-9B-A4B", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/HCHs/RivetCoder-9B-A4B
- SGLang
How to use HCHs/RivetCoder-9B-A4B 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 "HCHs/RivetCoder-9B-A4B" \ --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": "HCHs/RivetCoder-9B-A4B", "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 "HCHs/RivetCoder-9B-A4B" \ --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": "HCHs/RivetCoder-9B-A4B", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use HCHs/RivetCoder-9B-A4B with Docker Model Runner:
docker model run hf.co/HCHs/RivetCoder-9B-A4B
| { | |
| "completed_at": "2026-08-27T13:04:49Z", | |
| "final_checkpoint": "<workspace>\\artifacts\\router-training-qwen-all30-fixed16-top4\\checkpoint-00000060", | |
| "final_expert_off_exactness": { | |
| "atol": 0.0, | |
| "exact_equal": true, | |
| "max_abs_diff": 0.0, | |
| "within_tolerance": true | |
| }, | |
| "final_metrics": { | |
| "coding_sampler_epoch": 2, | |
| "cuda": { | |
| "allocated_bytes": 5474311680, | |
| "max_allocated_bytes": 12469938176, | |
| "reserved_bytes": 14224982016 | |
| }, | |
| "expert_off_exactness": { | |
| "atol": 0.0, | |
| "exact_equal": true, | |
| "max_abs_diff": 0.0, | |
| "within_tolerance": true | |
| }, | |
| "generic_sampler_epoch": 7, | |
| "global_step": 60, | |
| "grad_norm": 0.015457727015018463, | |
| "learning_rate": 0.0005, | |
| "learning_rates": { | |
| "residual_scale": 0.05, | |
| "router": 0.0005, | |
| "token_gate": 0.001 | |
| }, | |
| "loss": { | |
| "coding_ce": 0.3947601616382599, | |
| "coding_gate_target": 0.40677282214164734, | |
| "generic_gate_sparsity": 0.036865234375, | |
| "generic_kl": 0.0010666541056707501, | |
| "router_load_balance": 0.9907996654510498, | |
| "total": 0.5061666369438171 | |
| }, | |
| "router_diagnostics": [ | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 123, | |
| 219, | |
| 189, | |
| 205, | |
| 175, | |
| 180, | |
| 187, | |
| 164, | |
| 101, | |
| 119, | |
| 234, | |
| 170, | |
| 203, | |
| 192, | |
| 163, | |
| 180 | |
| ], | |
| "layer": 0, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 7.45758370612748e-05, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.058837890625, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 140, | |
| 179, | |
| 181, | |
| 170, | |
| 138, | |
| 170, | |
| 192, | |
| 253, | |
| 155, | |
| 144, | |
| 213, | |
| 204, | |
| 200, | |
| 206, | |
| 180, | |
| 79 | |
| ], | |
| "layer": 1, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.07198415696620941, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.04345703125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 167, | |
| 138, | |
| 202, | |
| 209, | |
| 134, | |
| 170, | |
| 192, | |
| 264, | |
| 174, | |
| 148, | |
| 136, | |
| 140, | |
| 167, | |
| 160, | |
| 190, | |
| 213 | |
| ], | |
| "layer": 2, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0014509361935779452, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0771484375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 251, | |
| 237, | |
| 206, | |
| 136, | |
| 146, | |
| 132, | |
| 170, | |
| 158, | |
| 191, | |
| 161, | |
| 0, | |
| 202, | |
| 157, | |
| 210, | |
| 247, | |
| 200 | |
| ], | |
| "layer": 3, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0868370309472084, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0615234375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 252, | |
| 167, | |
| 150, | |
| 191, | |
| 235, | |
| 182, | |
| 134, | |
| 0, | |
| 198, | |
| 244, | |
| 181, | |
| 201, | |
| 112, | |
| 190, | |
| 175, | |
| 192 | |
| ], | |
| "layer": 4, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08832801133394241, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.04736328125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 178, | |
| 129, | |
| 217, | |
| 126, | |
| 143, | |
| 192, | |
| 171, | |
| 220, | |
| 189, | |
| 245, | |
| 133, | |
| 391, | |
| 0, | |
| 149, | |
| 186, | |
| 135 | |
| ], | |
| "layer": 5, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.014818619005382061, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.056884765625, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 128, | |
| 128, | |
| 238, | |
| 194, | |
| 257, | |
| 141, | |
| 125, | |
| 181, | |
| 218, | |
| 278, | |
| 172, | |
| 238, | |
| 0, | |
| 121, | |
| 225, | |
| 160 | |
| ], | |
| "layer": 6, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08805593848228455, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.052001953125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 216, | |
| 145, | |
| 148, | |
| 169, | |
| 173, | |
| 194, | |
| 141, | |
| 157, | |
| 186, | |
| 203, | |
| 22, | |
| 277, | |
| 197, | |
| 264, | |
| 149, | |
| 163 | |
| ], | |
| "layer": 7, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.06830453872680664, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.048095703125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 180, | |
| 194, | |
| 248, | |
| 207, | |
| 180, | |
| 276, | |
| 184, | |
| 113, | |
| 248, | |
| 195, | |
| 124, | |
| 134, | |
| 125, | |
| 20, | |
| 132, | |
| 244 | |
| ], | |
| "layer": 8, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09049388021230698, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.046875, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 346, | |
| 292, | |
| 124, | |
| 226, | |
| 198, | |
| 262, | |
| 148, | |
| 256, | |
| 90, | |
| 34, | |
| 147, | |
| 189, | |
| 0, | |
| 180, | |
| 218, | |
| 94 | |
| ], | |
| "layer": 9, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09548121690750122, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.060302734375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 238, | |
| 178, | |
| 190, | |
| 61, | |
| 218, | |
| 152, | |
| 158, | |
| 143, | |
| 253, | |
| 190, | |
| 0, | |
| 191, | |
| 235, | |
| 154, | |
| 406, | |
| 37 | |
| ], | |
| "layer": 10, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09226511418819427, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.047607421875, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 218, | |
| 158, | |
| 16, | |
| 167, | |
| 229, | |
| 176, | |
| 0, | |
| 93, | |
| 157, | |
| 186, | |
| 182, | |
| 203, | |
| 262, | |
| 210, | |
| 338, | |
| 209 | |
| ], | |
| "layer": 11, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08412390202283859, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.044677734375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 238, | |
| 195, | |
| 242, | |
| 203, | |
| 303, | |
| 217, | |
| 73, | |
| 0, | |
| 36, | |
| 193, | |
| 144, | |
| 95, | |
| 222, | |
| 341, | |
| 147, | |
| 155 | |
| ], | |
| "layer": 12, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09192470461130142, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.047607421875, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 242, | |
| 176, | |
| 1, | |
| 203, | |
| 176, | |
| 304, | |
| 216, | |
| 165, | |
| 112, | |
| 145, | |
| 130, | |
| 126, | |
| 139, | |
| 201, | |
| 149, | |
| 319 | |
| ], | |
| "layer": 13, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0032708614598959684, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.06396484375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 192, | |
| 161, | |
| 272, | |
| 219, | |
| 1, | |
| 219, | |
| 118, | |
| 279, | |
| 197, | |
| 67, | |
| 197, | |
| 85, | |
| 168, | |
| 160, | |
| 267, | |
| 202 | |
| ], | |
| "layer": 14, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08825033158063889, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.05908203125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 180, | |
| 106, | |
| 257, | |
| 221, | |
| 243, | |
| 186, | |
| 296, | |
| 167, | |
| 0, | |
| 191, | |
| 157, | |
| 150, | |
| 201, | |
| 122, | |
| 173, | |
| 154 | |
| ], | |
| "layer": 15, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.08818262815475464, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.059814453125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 133, | |
| 188, | |
| 234, | |
| 223, | |
| 141, | |
| 174, | |
| 152, | |
| 165, | |
| 162, | |
| 159, | |
| 119, | |
| 166, | |
| 186, | |
| 216, | |
| 208, | |
| 178 | |
| ], | |
| "layer": 16, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0897134467959404, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0693359375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 211, | |
| 156, | |
| 168, | |
| 183, | |
| 259, | |
| 163, | |
| 154, | |
| 155, | |
| 130, | |
| 164, | |
| 128, | |
| 352, | |
| 215, | |
| 36, | |
| 169, | |
| 161 | |
| ], | |
| "layer": 17, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0917714461684227, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.16015625, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 163, | |
| 141, | |
| 173, | |
| 157, | |
| 170, | |
| 142, | |
| 247, | |
| 199, | |
| 144, | |
| 127, | |
| 218, | |
| 147, | |
| 199, | |
| 210, | |
| 114, | |
| 253 | |
| ], | |
| "layer": 18, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09746404737234116, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0057061342522501945, | |
| "token_gate_mean": 0.1787109375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 140, | |
| 242, | |
| 192, | |
| 123, | |
| 218, | |
| 101, | |
| 138, | |
| 230, | |
| 179, | |
| 265, | |
| 265, | |
| 137, | |
| 81, | |
| 205, | |
| 148, | |
| 140 | |
| ], | |
| "layer": 19, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09797238558530807, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.03423680365085602, | |
| "token_gate_mean": 0.2021484375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 103, | |
| 123, | |
| 141, | |
| 152, | |
| 173, | |
| 135, | |
| 244, | |
| 171, | |
| 176, | |
| 146, | |
| 216, | |
| 249, | |
| 129, | |
| 321, | |
| 177, | |
| 148 | |
| ], | |
| "layer": 20, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09263704717159271, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.08131241053342819, | |
| "token_gate_mean": 0.234375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 124, | |
| 268, | |
| 158, | |
| 85, | |
| 192, | |
| 181, | |
| 220, | |
| 101, | |
| 156, | |
| 254, | |
| 191, | |
| 171, | |
| 92, | |
| 148, | |
| 152, | |
| 311 | |
| ], | |
| "layer": 21, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09646926075220108, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.28388017416000366, | |
| "token_gate_mean": 0.396484375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 175, | |
| 179, | |
| 222, | |
| 147, | |
| 304, | |
| 237, | |
| 182, | |
| 156, | |
| 212, | |
| 214, | |
| 180, | |
| 72, | |
| 144, | |
| 137, | |
| 63, | |
| 180 | |
| ], | |
| "layer": 22, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09666198492050171, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.38944366574287415, | |
| "token_gate_mean": 0.439453125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 228, | |
| 196, | |
| 196, | |
| 128, | |
| 168, | |
| 106, | |
| 146, | |
| 271, | |
| 140, | |
| 220, | |
| 153, | |
| 135, | |
| 208, | |
| 99, | |
| 193, | |
| 217 | |
| ], | |
| "layer": 23, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08689017593860626, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.5049929022789001, | |
| "token_gate_mean": 0.4921875, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 233, | |
| 183, | |
| 123, | |
| 161, | |
| 133, | |
| 251, | |
| 236, | |
| 81, | |
| 290, | |
| 120, | |
| 213, | |
| 134, | |
| 212, | |
| 155, | |
| 125, | |
| 154 | |
| ], | |
| "layer": 24, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09673594683408737, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.48074179887771606, | |
| "token_gate_mean": 0.474609375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 284, | |
| 185, | |
| 166, | |
| 123, | |
| 115, | |
| 166, | |
| 175, | |
| 136, | |
| 297, | |
| 140, | |
| 278, | |
| 217, | |
| 90, | |
| 200, | |
| 140, | |
| 92 | |
| ], | |
| "layer": 25, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09465997666120529, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.48644793033599854, | |
| "token_gate_mean": 0.462890625, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 294, | |
| 195, | |
| 80, | |
| 127, | |
| 89, | |
| 170, | |
| 172, | |
| 139, | |
| 243, | |
| 279, | |
| 145, | |
| 241, | |
| 169, | |
| 106, | |
| 176, | |
| 179 | |
| ], | |
| "layer": 26, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09768165647983551, | |
| "router_entropy": 2.671875, | |
| "token_gate_active_fraction": 0.47931528091430664, | |
| "token_gate_mean": 0.4609375, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 118, | |
| 116, | |
| 190, | |
| 217, | |
| 242, | |
| 166, | |
| 184, | |
| 168, | |
| 100, | |
| 208, | |
| 184, | |
| 213, | |
| 138, | |
| 219, | |
| 143, | |
| 198 | |
| ], | |
| "layer": 27, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0952037125825882, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.5606276988983154, | |
| "token_gate_mean": 0.53515625, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 147, | |
| 258, | |
| 176, | |
| 108, | |
| 151, | |
| 184, | |
| 228, | |
| 93, | |
| 95, | |
| 101, | |
| 217, | |
| 183, | |
| 146, | |
| 307, | |
| 157, | |
| 253 | |
| ], | |
| "layer": 28, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09579590708017349, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.8388017416000366, | |
| "token_gate_mean": 0.73828125, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "coding", | |
| "expert_counts": [ | |
| 208, | |
| 134, | |
| 183, | |
| 56, | |
| 379, | |
| 296, | |
| 181, | |
| 169, | |
| 212, | |
| 73, | |
| 255, | |
| 172, | |
| 80, | |
| 75, | |
| 199, | |
| 132 | |
| ], | |
| "layer": 29, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09640846401453018, | |
| "router_entropy": 2.625, | |
| "token_gate_active_fraction": 0.9358060359954834, | |
| "token_gate_mean": 0.796875, | |
| "valid_tokens": 701 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 52, | |
| 76, | |
| 54, | |
| 68, | |
| 59, | |
| 76, | |
| 76, | |
| 59, | |
| 40, | |
| 51, | |
| 65, | |
| 49, | |
| 74, | |
| 56, | |
| 49, | |
| 60 | |
| ], | |
| "layer": 0, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 7.45758370612748e-05, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.034912109375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 46, | |
| 48, | |
| 71, | |
| 59, | |
| 57, | |
| 65, | |
| 64, | |
| 86, | |
| 51, | |
| 48, | |
| 71, | |
| 79, | |
| 66, | |
| 74, | |
| 65, | |
| 14 | |
| ], | |
| "layer": 1, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.07198415696620941, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0242919921875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 70, | |
| 61, | |
| 53, | |
| 52, | |
| 52, | |
| 60, | |
| 68, | |
| 107, | |
| 75, | |
| 15, | |
| 57, | |
| 65, | |
| 55, | |
| 57, | |
| 56, | |
| 61 | |
| ], | |
| "layer": 2, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0014509361935779452, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0260009765625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 60, | |
| 94, | |
| 63, | |
| 90, | |
| 41, | |
| 56, | |
| 65, | |
| 59, | |
| 59, | |
| 52, | |
| 0, | |
| 56, | |
| 65, | |
| 58, | |
| 63, | |
| 83 | |
| ], | |
| "layer": 3, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0868370309472084, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0224609375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 63, | |
| 67, | |
| 60, | |
| 67, | |
| 59, | |
| 65, | |
| 66, | |
| 0, | |
| 62, | |
| 76, | |
| 56, | |
| 51, | |
| 63, | |
| 58, | |
| 76, | |
| 75 | |
| ], | |
| "layer": 4, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08832801133394241, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.022216796875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 58, | |
| 62, | |
| 64, | |
| 58, | |
| 62, | |
| 75, | |
| 66, | |
| 50, | |
| 61, | |
| 55, | |
| 60, | |
| 114, | |
| 0, | |
| 61, | |
| 62, | |
| 56 | |
| ], | |
| "layer": 5, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.014818619005382061, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0230712890625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 51, | |
| 66, | |
| 72, | |
| 55, | |
| 81, | |
| 53, | |
| 67, | |
| 59, | |
| 67, | |
| 71, | |
| 53, | |
| 99, | |
| 0, | |
| 59, | |
| 52, | |
| 59 | |
| ], | |
| "layer": 6, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08805593848228455, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0238037109375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 61, | |
| 53, | |
| 67, | |
| 56, | |
| 63, | |
| 56, | |
| 44, | |
| 95, | |
| 64, | |
| 57, | |
| 22, | |
| 101, | |
| 60, | |
| 59, | |
| 47, | |
| 59 | |
| ], | |
| "layer": 7, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.06830453872680664, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0230712890625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 59, | |
| 69, | |
| 126, | |
| 62, | |
| 58, | |
| 60, | |
| 54, | |
| 51, | |
| 65, | |
| 76, | |
| 46, | |
| 58, | |
| 59, | |
| 9, | |
| 55, | |
| 57 | |
| ], | |
| "layer": 8, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09049388021230698, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0233154296875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 69, | |
| 66, | |
| 66, | |
| 72, | |
| 56, | |
| 75, | |
| 73, | |
| 63, | |
| 66, | |
| 6, | |
| 77, | |
| 67, | |
| 1, | |
| 67, | |
| 77, | |
| 63 | |
| ], | |
| "layer": 9, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09548121690750122, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.024658203125, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 60, | |
| 72, | |
| 63, | |
| 38, | |
| 60, | |
| 61, | |
| 57, | |
| 105, | |
| 65, | |
| 56, | |
| 0, | |
| 66, | |
| 66, | |
| 65, | |
| 88, | |
| 42 | |
| ], | |
| "layer": 10, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09226511418819427, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.02197265625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 73, | |
| 65, | |
| 51, | |
| 49, | |
| 68, | |
| 68, | |
| 0, | |
| 22, | |
| 50, | |
| 110, | |
| 64, | |
| 75, | |
| 61, | |
| 68, | |
| 74, | |
| 66 | |
| ], | |
| "layer": 11, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08412390202283859, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0213623046875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 63, | |
| 55, | |
| 75, | |
| 67, | |
| 66, | |
| 73, | |
| 21, | |
| 0, | |
| 49, | |
| 61, | |
| 63, | |
| 86, | |
| 69, | |
| 86, | |
| 65, | |
| 65 | |
| ], | |
| "layer": 12, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09192470461130142, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.023681640625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 76, | |
| 34, | |
| 5, | |
| 44, | |
| 58, | |
| 103, | |
| 64, | |
| 59, | |
| 47, | |
| 50, | |
| 56, | |
| 91, | |
| 64, | |
| 77, | |
| 71, | |
| 65 | |
| ], | |
| "layer": 13, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0032708614598959684, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0242919921875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 77, | |
| 50, | |
| 61, | |
| 71, | |
| 31, | |
| 42, | |
| 62, | |
| 57, | |
| 62, | |
| 54, | |
| 58, | |
| 59, | |
| 69, | |
| 67, | |
| 76, | |
| 68 | |
| ], | |
| "layer": 14, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08825033158063889, | |
| "router_entropy": 2.75, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.022705078125, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 61, | |
| 65, | |
| 93, | |
| 58, | |
| 58, | |
| 67, | |
| 65, | |
| 60, | |
| 0, | |
| 63, | |
| 38, | |
| 59, | |
| 60, | |
| 65, | |
| 60, | |
| 92 | |
| ], | |
| "layer": 15, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.08818262815475464, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0240478515625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 57, | |
| 94, | |
| 66, | |
| 48, | |
| 51, | |
| 65, | |
| 47, | |
| 66, | |
| 47, | |
| 103, | |
| 54, | |
| 50, | |
| 51, | |
| 59, | |
| 57, | |
| 49 | |
| ], | |
| "layer": 16, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0897134467959404, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0264892578125, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 58, | |
| 74, | |
| 41, | |
| 64, | |
| 60, | |
| 64, | |
| 49, | |
| 65, | |
| 48, | |
| 60, | |
| 78, | |
| 65, | |
| 65, | |
| 31, | |
| 72, | |
| 70 | |
| ], | |
| "layer": 17, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.0917714461684227, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0380859375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 65, | |
| 51, | |
| 76, | |
| 77, | |
| 61, | |
| 48, | |
| 59, | |
| 61, | |
| 46, | |
| 47, | |
| 55, | |
| 120, | |
| 60, | |
| 34, | |
| 55, | |
| 49 | |
| ], | |
| "layer": 18, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09746404737234116, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.035888671875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 49, | |
| 58, | |
| 64, | |
| 38, | |
| 61, | |
| 67, | |
| 79, | |
| 65, | |
| 71, | |
| 52, | |
| 56, | |
| 51, | |
| 53, | |
| 85, | |
| 81, | |
| 34 | |
| ], | |
| "layer": 19, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09797238558530807, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.03857421875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 61, | |
| 87, | |
| 68, | |
| 63, | |
| 56, | |
| 61, | |
| 61, | |
| 59, | |
| 58, | |
| 61, | |
| 33, | |
| 75, | |
| 51, | |
| 67, | |
| 44, | |
| 59 | |
| ], | |
| "layer": 20, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09263704717159271, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.043212890625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 43, | |
| 92, | |
| 66, | |
| 61, | |
| 53, | |
| 49, | |
| 62, | |
| 61, | |
| 53, | |
| 69, | |
| 73, | |
| 56, | |
| 38, | |
| 61, | |
| 51, | |
| 76 | |
| ], | |
| "layer": 21, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09646926075220108, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.04833984375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 64, | |
| 59, | |
| 72, | |
| 63, | |
| 55, | |
| 60, | |
| 61, | |
| 55, | |
| 74, | |
| 61, | |
| 56, | |
| 25, | |
| 62, | |
| 86, | |
| 41, | |
| 70 | |
| ], | |
| "layer": 22, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09666198492050171, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.047607421875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 64, | |
| 64, | |
| 54, | |
| 68, | |
| 48, | |
| 62, | |
| 65, | |
| 57, | |
| 63, | |
| 70, | |
| 70, | |
| 41, | |
| 62, | |
| 68, | |
| 50, | |
| 58 | |
| ], | |
| "layer": 23, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.08689017593860626, | |
| "router_entropy": 2.703125, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.04833984375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 53, | |
| 67, | |
| 62, | |
| 65, | |
| 58, | |
| 57, | |
| 59, | |
| 66, | |
| 61, | |
| 52, | |
| 55, | |
| 62, | |
| 57, | |
| 59, | |
| 76, | |
| 55 | |
| ], | |
| "layer": 24, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09673594683408737, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.0322265625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 56, | |
| 54, | |
| 63, | |
| 57, | |
| 63, | |
| 58, | |
| 43, | |
| 78, | |
| 54, | |
| 63, | |
| 61, | |
| 56, | |
| 58, | |
| 75, | |
| 53, | |
| 72 | |
| ], | |
| "layer": 25, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09465997666120529, | |
| "router_entropy": 2.71875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.031494140625, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 69, | |
| 58, | |
| 63, | |
| 70, | |
| 61, | |
| 51, | |
| 60, | |
| 61, | |
| 62, | |
| 49, | |
| 73, | |
| 59, | |
| 57, | |
| 56, | |
| 52, | |
| 63 | |
| ], | |
| "layer": 26, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09768165647983551, | |
| "router_entropy": 2.671875, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.03125, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 56, | |
| 60, | |
| 51, | |
| 71, | |
| 72, | |
| 75, | |
| 59, | |
| 46, | |
| 43, | |
| 65, | |
| 60, | |
| 70, | |
| 75, | |
| 50, | |
| 43, | |
| 68 | |
| ], | |
| "layer": 27, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.0952037125825882, | |
| "router_entropy": 2.734375, | |
| "token_gate_active_fraction": 0.0, | |
| "token_gate_mean": 0.05029296875, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 49, | |
| 76, | |
| 43, | |
| 71, | |
| 76, | |
| 78, | |
| 56, | |
| 63, | |
| 58, | |
| 51, | |
| 55, | |
| 61, | |
| 75, | |
| 52, | |
| 39, | |
| 61 | |
| ], | |
| "layer": 28, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": -0.09579590708017349, | |
| "router_entropy": 2.6875, | |
| "token_gate_active_fraction": 0.012448133900761604, | |
| "token_gate_mean": 0.10693359375, | |
| "valid_tokens": 241 | |
| }, | |
| { | |
| "domain": "generic", | |
| "expert_counts": [ | |
| 83, | |
| 35, | |
| 60, | |
| 52, | |
| 76, | |
| 74, | |
| 83, | |
| 57, | |
| 73, | |
| 54, | |
| 44, | |
| 63, | |
| 73, | |
| 49, | |
| 39, | |
| 49 | |
| ], | |
| "layer": 29, | |
| "mean_selected_weight": 0.25, | |
| "residual_scale": 0.09640846401453018, | |
| "router_entropy": 2.640625, | |
| "token_gate_active_fraction": 0.0456431545317173, | |
| "token_gate_mean": 0.1435546875, | |
| "valid_tokens": 241 | |
| } | |
| ], | |
| "timestamp": "2026-08-27T13:02:49Z" | |
| }, | |
| "global_step": 60, | |
| "run_fingerprint": "68f9fa90e18e180f7037dea5e057aa296c95b59267aa09f38285363b5e517c2a", | |
| "validation": { | |
| "ce_improvement": 0.0043891161448869065, | |
| "finite": true, | |
| "fused_ce": 0.6014361901441425, | |
| "host_expert_off_ce": 0.6058253062890294, | |
| "mean_absolute_residual_scale": 0.07969798704336123, | |
| "mean_residual_scale": -0.002938792503376438, | |
| "mean_token_gate": 0.19380086263020832, | |
| "records": 4, | |
| "supervised_tokens": 1417 | |
| } | |
| } | |