Upload folder using huggingface_hub

Dockerfile

@@ -1,25 +1,4 @@
-# Stage 1: Merge LoRA adapter into base model (CPU only)
-FROM python:3.11-slim AS merger
-
-RUN pip install --no-cache-dir \
-    torch==2.5.1 --index-url https://download.pytorch.org/whl/cpu \
-    && pip install --no-cache-dir \
-    transformers==4.51.3 \
-    peft==0.18.1 \
-    safetensors \
-    accelerate
-
-COPY adapter_config.json /adapter/
-COPY adapter_model.safetensors /adapter/
-COPY tokenizer.json /adapter/
-COPY tokenizer_config.json /adapter/
-
-COPY merge_and_save.py /merge_and_save.py
-
-RUN python /merge_and_save.py
-
-# Stage 2: Lightweight runtime (CPU — free tier on HF Spaces)
-FROM python:3.11-slim AS runtime
+FROM python:3.11-slim
 
 RUN pip install --no-cache-dir \
     torch==2.5.1 --index-url https://download.pytorch.org/whl/cpu \
@@ -32,11 +11,10 @@ RUN pip install --no-cache-dir \
 
 RUN useradd -m -u 1000 appuser
 
-COPY --from=merger --chown=1000:1000 /merged_model /model
 COPY --chown=1000:1000 server.py /app/server.py
 
 USER appuser
-ENV MODEL_DIR=/model
+ENV MODEL_DIR=nmysore/seng-model
 EXPOSE 7860
 
 CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860", "--app-dir", "/app"]

adapter_config.json

@@ -1,41 +0,0 @@
-{
-  "alora_invocation_tokens": null,
-  "alpha_pattern": {},
-  "arrow_config": null,
-  "auto_mapping": null,
-  "base_model_name_or_path": "gpt2-large",
-  "bias": "none",
-  "corda_config": null,
-  "ensure_weight_tying": false,
-  "eva_config": null,
-  "exclude_modules": null,
-  "fan_in_fan_out": true,
-  "inference_mode": true,
-  "init_lora_weights": true,
-  "layer_replication": null,
-  "layers_pattern": null,
-  "layers_to_transform": null,
-  "loftq_config": {},
-  "lora_alpha": 32,
-  "lora_bias": false,
-  "lora_dropout": 0.05,
-  "megatron_config": null,
-  "megatron_core": "megatron.core",
-  "modules_to_save": null,
-  "peft_type": "LORA",
-  "peft_version": "0.18.1",
-  "qalora_group_size": 16,
-  "r": 16,
-  "rank_pattern": {},
-  "revision": null,
-  "target_modules": [
-    "c_proj",
-    "c_attn"
-  ],
-  "target_parameters": null,
-  "task_type": "CAUSAL_LM",
-  "trainable_token_indices": null,
-  "use_dora": false,
-  "use_qalora": false,
-  "use_rslora": false
-}

adapter_model.safetensors

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:3be1c1ea2eba9b7ba1303c969b4ba41dcc6cb1fc2bb1aeec66653167d904745a
-size 32468368

merge_and_save.py

@@ -1,38 +0,0 @@
-"""Merge LoRA adapter into gpt2-large base model and save.
-
-Runs at Docker build time (CPU only). The merged model eliminates the
-PEFT dependency at runtime, simplifying the serving image.
-"""
-
-import torch
-from peft import PeftModel
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-ADAPTER_DIR = "/adapter"
-OUTPUT_DIR = "/merged_model"
-BASE_MODEL = "gpt2-large"
-
-
-def main():
-    print(f"Loading base model: {BASE_MODEL}")
-    base_model = AutoModelForCausalLM.from_pretrained(
-        BASE_MODEL, torch_dtype=torch.float32
-    )
-
-    print(f"Loading LoRA adapter from: {ADAPTER_DIR}")
-    model = PeftModel.from_pretrained(base_model, ADAPTER_DIR)
-
-    print("Merging adapter into base model...")
-    model = model.merge_and_unload()
-
-    print(f"Saving merged model to: {OUTPUT_DIR}")
-    model.save_pretrained(OUTPUT_DIR)
-
-    tokenizer = AutoTokenizer.from_pretrained(ADAPTER_DIR)
-    tokenizer.save_pretrained(OUTPUT_DIR)
-
-    print("Merge complete.")
-
-
-if __name__ == "__main__":
-    main()

tokenizer_config.json

@@ -1,12 +0,0 @@
-{
-  "add_prefix_space": false,
-  "backend": "tokenizers",
-  "bos_token": "<|endoftext|>",
-  "eos_token": "<|endoftext|>",
-  "errors": "replace",
-  "is_local": false,
-  "model_max_length": 1024,
-  "pad_token": "<|endoftext|>",
-  "tokenizer_class": "GPT2Tokenizer",
-  "unk_token": "<|endoftext|>"
-}