"""Installed gate/up row-map oracle, using the identical native fixture inputs.""" import hashlib,json,subprocess,sys from pathlib import Path checkout,root,native=map(Path,sys.argv[1:4]) assert subprocess.check_output(['git','-C',str(checkout),'rev-parse','HEAD'],text=True).strip()=='9882703f3105363ddc37eca9f97aa09a1d387112' sys.path.insert(0,str(checkout)) import mlx.core as mx import numpy as np from mtplx.kernels import moe_sorted_gather_nax as kernel assert mx.__version__=='0.32.2' index=json.loads((root/'model.safetensors.index.json').read_text())['weight_map'] base='language_model.model.layers.0.mlp.switch_mlp.' weights=mx.load(str(root/index[base+'gate_proj.weight'])) def parts(proj):return [weights[base+proj+'.'+part] for part in ['weight','scales','biases']] gate=parts('gate_proj');up=parts('up_proj') gu=[mx.concatenate(a,axis=1) for a in zip(gate,up)] mx.eval(*gu,*gate,*up) def digest(a): mx.eval(a) return hashlib.sha256(np.asarray(a.view(mx.uint16)).tobytes()).hexdigest() count=0 for line in native.read_text().splitlines(): if '"case":"installed_gate_up"' not in line:continue case=json.loads(line[line.index('{'):]);count+=1 tokens,rows=case['tokens'],case['rows'] v=((np.arange(tokens*2560,dtype=np.uint32)*17+7*13)%257).astype(np.int32)-128 x=mx.array((v.astype(np.float32)/128).reshape(tokens,1,2560)).astype(mx.bfloat16) row_map=mx.arange(rows,dtype=mx.uint32)%tokens ids=mx.array(np.arange(rows,dtype=np.uint32)*512//rows) g,u=(gate,up) if case['separate'] else (gu,None) kw=dict(group_size=64,bits=4,swiglu=True,up=u) y=kernel.gather_rows_qmm(x,row_map,*g,ids,**kw) stock=kernel._stock(x,row_map,*g,ids,**kw) assert y is not None and digest(y)==digest(stock)==case['sha256'],case print(json.dumps(dict(case='installed_oracle',native=case,exact=True,kernel=kernel.stats())),flush=True) del x,row_map,ids,y,stock mx.clear_cache() assert count==12,count