NL→MLIR benchmarks (465 instances; arith/func/memref, linalg, StableHLO). Code: github.com/plawanrath/slm-to-mlir-constrained-emitter
Plawan Kumar Rath PRO
plawanrath
AI & ML interests
None yet
Recent Activity
updated a dataset 1 day ago
plawanrath/MLIR-Functional-Reference-30 updated a dataset 1 day ago
plawanrath/StableHLO-OutOfGrammar-25 updated a dataset 1 day ago
plawanrath/StableHLO-Held-Out-200Organizations
None yet
Quantization Undoes Alignment — IEEE Cloud Summit 2026
MLX artifacts for the IEEE Cloud Summit 2026 paper. 3 LLMs × 5 precisions (BF16/Q8/Q6/Q4/Q3, gs=64).
-
plawanrath/phi-3.5-mini-instruct-q3-mlx-cba
Text Generation • 4B • Updated • 33 -
plawanrath/phi-3.5-mini-instruct-q4-mlx-cba
Text Generation • 4B • Updated • 36 -
plawanrath/phi-3.5-mini-instruct-q6-mlx-cba
Text Generation • 4B • Updated • 28 -
plawanrath/mistral-7b-instruct-v0.3-q3-mlx-cba
Text Generation • 7B • Updated • 24
Weight Pruning Amplifies Bias (AIIoT 2026)
36 pruned-LLM artifacts (3 models x 3 methods x 4 sparsities) for AIIoT 2026 bias-amplification paper. Research only - not for deployment.
-
plawanrath/gemma-2-9b-it-magnitude-s10-pia
Text Generation • 9B • Updated • 14 -
plawanrath/gemma-2-9b-it-magnitude-s30-pia
Text Generation • 9B • Updated • 6 -
plawanrath/gemma-2-9b-it-magnitude-s50-pia
Text Generation • 9B • Updated • 9 -
plawanrath/gemma-2-9b-it-magnitude-s70-pia
Text Generation • 9B • Updated • 11
Cross-Dialect MLIR Benchmarks - NeurIPS 2026 (E&D)
NL→MLIR benchmarks (465 instances; arith/func/memref, linalg, StableHLO). Code: github.com/plawanrath/slm-to-mlir-constrained-emitter
Weight Pruning Amplifies Bias (AIIoT 2026)
36 pruned-LLM artifacts (3 models x 3 methods x 4 sparsities) for AIIoT 2026 bias-amplification paper. Research only - not for deployment.
-
plawanrath/gemma-2-9b-it-magnitude-s10-pia
Text Generation • 9B • Updated • 14 -
plawanrath/gemma-2-9b-it-magnitude-s30-pia
Text Generation • 9B • Updated • 6 -
plawanrath/gemma-2-9b-it-magnitude-s50-pia
Text Generation • 9B • Updated • 9 -
plawanrath/gemma-2-9b-it-magnitude-s70-pia
Text Generation • 9B • Updated • 11
Quantization Undoes Alignment — IEEE Cloud Summit 2026
MLX artifacts for the IEEE Cloud Summit 2026 paper. 3 LLMs × 5 precisions (BF16/Q8/Q6/Q4/Q3, gs=64).
-
plawanrath/phi-3.5-mini-instruct-q3-mlx-cba
Text Generation • 4B • Updated • 33 -
plawanrath/phi-3.5-mini-instruct-q4-mlx-cba
Text Generation • 4B • Updated • 36 -
plawanrath/phi-3.5-mini-instruct-q6-mlx-cba
Text Generation • 4B • Updated • 28 -
plawanrath/mistral-7b-instruct-v0.3-q3-mlx-cba
Text Generation • 7B • Updated • 24