I am getting a compilation error while fine tuning the model likely an environment or library compatibility issue

[ChPuru]

---

CompilationError Traceback (most recent call last)
/tmp/ipython-input-744766663.py in <cell line: 0>()
119
120 print(f"--- Starting Training on {torch.cuda.device_count()} GPUs ---")
--> 121 trainer.train()
122 print("--- Training Complete ---")
123

<>33 frames

/usr/local/lib/python3.12/dist-packages/triton/compiler/compiler.py in make_ir(self, options, codegen_fns, module_map, context)

CompilationError: at 71:12:
v_ptrs = V + off_b * stride_vb + off_h * stride_vh + (
offs_n[:, None] * stride_vn + offs_d[None, :])
if BIAS_TYPE == 'vector':
b_ptrs = Bias + off_b * stride_bb + off_h * stride_bh + offs_n
elif BIAS_TYPE == 'matrix':
b_ptrs = Bias + off_b * stride_bb + off_h * stride_bh + (
offs_m[:, None] * stride_bm + offs_n[None, :])
else:
raise ValueError("BIAS_TYPE must be one of {'vector', 'matrix'}")
# initialize pointer to m and l
t_ptrs = TMP + off_hb * seqlen_q_rounded + offs_m
lse_i = tl.zeros([BLOCK_M], dtype=tl.float32) - float('inf')
^
OSError('could not get source code')