Remove gpt_oss test code and add in examples

Signed-off-by: Sharif Inamdar <[email protected]>

Author: isharif168

src/llmcompressor/modeling/gpt_oss.py

@@ -172,45 +172,3 @@ def forward(self, hidden_states):
         router_scores = router_scores.view(B * T, -1)  # shape doesn't matter much; it’s ignored by the decoder
         return out, router_scores
 
-
-model_id = "unsloth/gpt-oss-120b-BF16"
-
-model = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    torch_dtype=torch.bfloat16,
-    device_map="auto",
-    trust_remote_code=True,
-)
-tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
-
-convert_model_for_quantization_gptoss(model)
-
-# -----------------------------
-# Quantization recipe
-# -----------------------------
-recipe = QuantizationModifier(
-    targets="Linear",
-    scheme="FP8_DYNAMIC",
-    ignore=[
-        "re:.*lm_head",
-        "re:.*self_attn",
-        "re:.*attn",
-        "re:.*attention.*",
-        "re:.*router",
-    ],
-)
-
-SAVE_DIR = f"{model_id.split('/')[-1]}-FP8-Dynamic"
-
-# Oneshot quantization
-oneshot(
-    model=model,
-    tokenizer=tokenizer,
-    recipe=recipe,
-    trust_remote_code_model=True,
-    output_dir=SAVE_DIR,
-)
-
-# Save compressed
-model.save_pretrained(SAVE_DIR, save_compressed=True)
-tokenizer.save_pretrained(SAVE_DIR)